Mansor, Rozaihan (2012) Proteomic and Metabolomic Studies on Milk During Bovine Mastitis

Mansor, Rozaihan (2012) Proteomic and metabolomic studies on milk during bovine mastitis.

PhD thesis

http://theses.gla.ac.uk/3207/

A copy can be downloaded for personal non-commercial research or study, without prior permission or charge

This thesis cannot be reproduced or quoted extensively from without first obtaining permission in writing from the Author

The content must not be changed in any way or sold commercially in any format or medium without the formal permission of the Author

When referring to this work, full bibliographic details including the author, title, awarding institution and date of the thesis must be given

Glasgow Theses Service http://theses.gla.ac.uk/ [email protected] 1

Proteomic and Metabolomic Studies on Milk during Bovine Mastitis

Rozaihan Mansor DVM

Submitted in fulfilment of the requirements for the degree of DOCTOR OF PHILOSOPHY

Institute of Infection Immunity and Inflammation Life Sciences College of Medical, Veterinary & Life Sciences University of Glasgow

February 2012

Abstract

The principal objectives of the study presented in this thesis were to study the changes of milk proteomes, peptidomes and metabolomes during the course of bovine mastitis in comparison with normal milk samples and to discover new bovine mastitis biomarkers using various modern and up-to-date methodologies such as proteomics, peptidomics and metabolomics.

Bovine mastitis caused by bacterial infection of the mammary gland of dairy cows is often associated with loss of milk production due to a reduction in milk composition and quality which in turns, lead to negative economic impact on dairy industry.

Two important acute phase proteins (APPs) which serve as valuable biomarkers in bovine mastitis were investigated in every chapter using developed and validated enzyme linked immunosorbent assay (ELISA) for bovine milk haptoglobin and commercially available ELISA for bovine milk serum amyloid A3 (M-SAA3). These APPs were quantified alongside somatic cell counts (SCC) and California Mastitis Test (CMT) to confirm the disease status of each animal used in this study.

Proteomic methodologies were applied including 1D gel electrophoresis, 2D gel electrophoresis, MALDI-TOF analysis and difference gel electrophoresis to investigate the changes of milk proteome in both subclinical and clinical mastitic milk samples in comparison with healthy milk samples. However these investigations did not reveal novel biomarkers for mastitis.

Next, peptidomic methodologies were used to study the changes in milk peptidome and to detect the presence of any significant disease biomarkers in the presence of bovine mastitis by using CE-MS and LC-MS/MS. A total of 31 and 14 polypeptides can be used to discriminate control from infected groups and E. coli from S. aureus infected groups respectively.

Lastly, metabolomic methodology was applied with an intention to study the changes in milk metabolome and ultimately to detect the presence of novel biomarkers in bovine mastitis. Di- and tri-peptides were found higher in S. aureus than in E. coli infected groups and based on metabolic pathways, arachidonic, arginine and galactose metabolites were seen increased in mastitic milk samples in comparison to healthy milk samples. 3

Overall, the findings detailed in this thesis indicate that the use of advanced proteomic and metabolomic methodologies could deliver on their promise of the discovery of potential significant bovine mastitis biomarkers. Further studies are needed for validation of these proposed biomarkers and it was hoped that better prevention and treatment methods for bovine mastitis can be achieved in the future.

Table of Content

Proteomic and Metabolomic Studies on Milk During Bovine Mastitis...... 1

Abstract...... 2

Table of Content...... 4

List of Tables...... 9

List of Figures...... 12

List of Abbreviations...... 14

Acknowledgements..………………………...... ………………………….……………..26

Author’s Declaration.……………………………………………………..…...... ……...28

Dedication.………………………………………………………………….…………...... 29

Chapter I Introduction..………………………………………………………………....30

1.1 Bovine Mastitis and its Impact on the Dairy Industry...... 30 1.1.1 Bovine Milk...... 31 1.1.2 Milk Protein...... 32 1.2 Diagnostic Methods for Bovine Mastitis...... 33 1.2.1 Somatic Cell Counts...... 33 1.2.2 California Mastitis Tes (CMT)...... 36 1.2.3 Enzyme levels...... 39 1.2.4 Electrical Conductivity...... 44 1.2.5 Infrared Thermography (IRT)...... 48 1.3 Biomarkers in bovine mastitis...... 49 1.4 Acute Phase Proteins in Bovine Mastitis...... 52 1.4.1 Naturally occurring mastitis...... 54 1.4.2 Experimental Induced Mastitis...... 56 1.4.3 Acute Phase Protein and Milk Quality...... 59 1.5 General Proteomics Application...... 60 1.5.1 Proteomics in Animal Sciences...... 61 1.5.2 Milk Proteomics...... 63 1.5.3 Proteomics on Bovine Mastitis...... 70 1.6 General Peptidomics...... 72 1.6.1 The presence of peptides in bovine milk and dairy products...... 74 1.6.2 Peptidomics in Bovine Milk/Bovine Mastitis...... 77 1.7 Milk Metabolomics...... 78 1.7.1 Metabolites found in bovine milk/serum...... 81 1.8 Aims of study...... 83

Chapter II Acute Phase Protein in Bovine Mastitis……………………………...... 84

2.1 Introduction...... 84 5

2.1.1 Bovine Acute Phase Proteins and Acute Phase Reactions...... 84 2.1.2 Bovine Haptoglobin...... 85 2.1.3 Haptoglobin in bovine mastitis...... 86 2.1.4 Quantification of bovine Hp...... 88 2.1.5 Bovine Serum Amyloid A (SAA)...... 91 2.1.6 Serum Amyloid A in Bovine Mastitis...... 92 2.1.7 Quantification of bovine SAA...... 96 2.1.8 Aims of Study...... 98 2.2 Materials and Methods...... 98 2.2.1 Reagents...... 98 2.2.2 Milk samples...... 98 2.2.3 California Mastitis Test...... 98 2.2.4 Development of Bovine Haptoglobin ELISA...... 98 2.2.4.1 Biotinylation of antibody...... 99 2.2.4.1.1 Preparation of Elution buffer and Antibody Binding Solution...... 99 2.2.4.1.2 Antibody Binding...... 99 2.2.4.1.3 Antibody Biotinylation...... 99 2.2.4.1.4 Antibody Elution...... 100 2.2.4.2 Newly prepared biotinylated antibody...... 100 2.2.4.3 Assay Procedure...... 101 2.2.4.3.1 Coating...... 101 2.2.4.3.2 Washing...... 101 2.2.4.3.3 Blocking...... 101 2.2.4.3.4 Washing...... 101 2.2.4.3.5 Dilution and addition of Standard Bovine Haptoglobin and Milk Samples...... 101 2.2.4.3.6 Addition of biotinylated antibody...... 101 2.2.4.3.7 Addition of secondary antibody...... 101 2.2.4.3.8 Addition of substrate...... 102 2.2.4.3.9 Absorbance reading...... 102 2.2.5 Haptoglobin Assay Optimisation...... 102 2.2.5.1 Antibody concentrations...... 102 2.2.5.2 Varying batch of biotinylated antibodies...... 102 2.2.5.3 Biotinylated antibody concentrations...... 102 2.2.5.4 Optimisation of Standard and Sample Dilution...... 103 2.2.6 Haptoglobin Assay Validation...... 103 2.2.6.1 Precision...... 103 2.2.6.2 Accuracy...... 103 2.2.6.3 Limit of detection...... 103 2.2.6.4 Specificity...... 103 2.2.7 Serum Amyloid A Assay...... 103 2.2.7.1 Reagents...... 104 2.2.7.2 Reagent Preparation...... 104 2.2.7.3 SAA Assay...... 104 2.2.8 SAA Assay Validation...... 104 6

2.2.8.1 SAA Assay Precision...... 104 2.2.8.2 Limit of Detection...... 105 2.2.9 Clinical Validation...... 105 2.2.10 Data Analysis...... 105 2.3 Results...... 105 2.3.1 California Mastitis Test...... 105 2.3.2 Immuno dot blot of newly prepared biotinylated antibody...... 105 2.3.3 Haptoglobin Assay Optimisation...... 105 2.3.3.1 Coating Antibody concentrations...... 105 2.3.3.2 Biotinylated antibody concentrations...... 106 2.3.3.3 Varying batch of biotinylated antibodies...... 106 2.3.3.4 Optimisation of standard and sample dilution...... 111 2.3.4 Haptoglobin Assay Validation...... 111 2.3.4.1 Precision...... 111 2.3.4.2 Accuracy...... 111 2.3.4.3 Limit of detection...... 111 2.3.4.4 Specificity...... 115 2.3.5 SAA Assay Validation...... 115 2.3.5.1 Precision...... 115 2.3.5.2 Limit of detection...... 115 2.3.6 Clinical Validation...... 115 2.4 Discussion...... 126 2.4.1 Clinical Validation Assesment of APP level in healthy, subclinical and clinical mastitic milk samples...... 127 2.4.2 Hp level in three different groups of milk samples...... 127 2.4.3 SAA level in 3 different groups of milk samples...... 128 2.4.4 Correlation between SAA and Hp levels in milk samples...... 128 2.4.5 Correlation between SCC and APP levels in milk samples...... 129

Chapter III Proteomics in Bovine Mastitis...... 131

3.1 Introduction...... 131 3.1.1 Sample Preparation and Collection...... 131 3.1.2 Protein Separation...... 132 3.1.3 Protein Characterization...... 134 3.1.4 Quantitative Proteomics...... 135 3.1.5 Aims of Study...... 135 3.2 Materials and Methods...... 136 3.2.1 Reagents...... 136 3.2.2 Milk samples...... 136 3.2.3 California Mastitis Test...... 136 3.2.4 Removal of high abundance milk proteins...... 136 3.2.4.1 Rennet Coagulation...... 136 3.2.4.2 Acid Precipitation...... 137 3.2.4.3 Ammonium sulphate (NH4SO4) Fractionation...... 137 3.2.5 One-dimensional gel electrophoresis...... 137 3.2.5.1 Reagent Preparation and Stock Solutions...... 137 7

3.2.5.2 Sample Preparation...... 137 3.2.5.3 Running the gel electrophoresis...... 138 3.2.5.4 Gel staining and Image Acquisition...... 138 3.2.5.5 In-gel trypsin digestion of Coomassie-stained proteins....138 3.2.5.6 Sample Plating...... 139 3.2.6 Two-dimensional gel electrophoresis...... 139 3.2.6.1 Sample preparation...... 139 3.2.6.2 Isoelectric Focusing...... 139 3.2.6.3 SDS-PAGE Gel Electrophoresis...... 140 3.2.6.4 Gel staining and Image Acquisition...... 140 3.2.7 Difference Gel Electrophoresis...... 140 3.2.7.1 Sample Preparation...... 141 3.2.7.2 CyDye Labelling...... 141 3.2.7.3 Isoelectric focussing...... 141 3.2.7.4 SDS-PAGE...... 141 3.2.7.5 Gel Scanning...... 142 3.2.7.6 Image manipulation...... 142 3.3 Results...... 142 3.3.1 SCC measurement and acute phase proteins...... 142 3.3.2 Comparison between rennet coagulation, acid precipitation and salt precipitation (ammonium sulphate) to remove casein...... 143 3.3.3 1D gel electrophoresis on individual subclinical mastitic and healthy milk samples...... 150 3.3.4 2D gel electrophoresis on pooled subclinical mastitic and healthy milk samples...... 150 3.3.5 Difference gel electrophoresis (DIGE) on pooled clinical mastitic and healthy milk samples...... 157 3.4 Discussion...... 161

Chapter IV Peptidomic Analysis on Bovine Mastitis...... 167

4.1 Introduction...... 167 4.1.1 Aims of Study...... 171 4.2 Materials and Methods...... 172 4.2.1 Milk Samples and Reagents...... 172 4.2.2 Milk sample preparation for CE-MS using PD-10 Desalting...... 172 4.2.2.1 Preparation of fresh ammonium (NH4OH) and urea buffer...... 172 4.2.2.2 Sample Preparation...... 172 4.2.3 Protein concentration...... 173 4.2.4 CE-MS Analysis...... 173 4.2.4.1 Preparation of buffers for CE-MS analysis...... 173 4.2.4.2 CE-MS Analysis for peptide fingerprint...... 174 4.2.5 CE-MS Data processing...... 174 4.2.6 Statistical analysis...... 175 4.2.7 LC-MS Analysis for Peptide Sequencing...... 175 4.2.7.1 LC-MS/MS analysis...... 175 8

4.2.7.2 LC-MS/MS- Data analysis...... 176 4.3 Results ...... 176 4.3.1 SCC measurement and acute phase proteins concentration...... 176 4.3.2 Polypeptide Fingerprinting for control and infected groups...... 181 4.3.3 Model Generation of the polypeptide fingerprints for control and infected groups...... 183 4.3.4 CE-MS Analysis...... 188 4.3.5 Peptide sequencing...... 193 4.3.6 Protein identification...... 194 4.4 Discussion...... 199

Chapter V Metabolomics Analysis in Bovine Matitis...... 205

5.1 General Metabolomics...... 205 5.1.1 Metabolites in milk and serum found during bovine mastitis...... 206 5.1.2 Aims of Study...... 212 5.2 Materials and Methods...... 212 5.2.1 Animals and Samples...... 212 5.2.2 Chloroform and methanol Extraction...... 212 5.2.2.1 Reagent preparation...... 212 5.2.2.2 Extraction method...... 213 5.2.3 LC-MS Analysis...... 213 5.2.4 Data Acquisition...... 213 5.2.5 Analysis of Individual Metabolites...... 213 5.2.6 Metabolites Analysis – Pathway Analysis...... 214 5.3 Results...... 215 5.3.1 SCC measurement and acute phase proteins concentration...... 215 5.3.2 Metabolites found in milk...... 219 5.3.3 Effect of mastitis on the metabolites of milk...... 221 5.3.4 Relative changes...... 237 5.3.5 Metabolic Pathways Altered by Bovine Mastitis...... 252 5.4 Discussion...... 262 5.4.1 Lipid Metabolism...... 263 5.4.2 Carbohydrate Metabolism...... 266 5.4.3 Protein Metabolism...... 268

Chapter VI General Discussion ...... 271 Appendices...... 278 List of References...... 282 9

List of Tables

Table 1-1: Examples of casein, whey and other proteins present in bovine milk composition...... 33

Table 1-2: Description on the CMT reaction of milk and their CMT scores...... 37

Table 2-1: Inter-assay precision results obtained for ELISA with milk samples...... 113

Table 2-3: The bacteria causing mastitis cultured from the clinical mastitic milk samples...... 120

Table 2-4: Spearman’s correlation between SAA concentration, Hp concentration and SCC measurements in milk samples...... 124

Table 2-5: Mann-Whitney’s Test for SAA concentration, Hp concentration and SCC measurement between healthy, subclinical mastitic and clinical mastitic groups...... 125

Table 3-1: The SCC measurement in all control and subclinical mastitic milk samples and CMT scores in subclinical mastitic milk samples used in this study...... 144

Table 3-2: The SCC measurement in all control milk samples and APP concentration in clinically infected milk samples used in this study for DiGE...... 146

Table 3-3: Proteins identified after in-gel trypsin digestion of protein abnds on 1D gel electrophoresis...... 154

Table 4-1: The acute phase proteins (SAA and Hp) concentrations and SCC measurement in all control milk samples used in this study...... 178

Table 4-2a: The acute phase proteins (SAA and Hp) concentrations and bacterial causing mastitis in infected milk samples caused by E. coli used in this study...... 179

Table 4-2b: The acute phase proteins (SAA and Hp) concentrations and bacterial causing mastitis in infected milk samples caused by S. aureus used in this study...... 180

Table 4-3: Polypeptides that were identified from 3 models generated to discriminate E. coli from S. aureus infected groups...... 186

Table 4-4: A total of 13 polypeptides generated from 3 models that discriminate E. coli from S. aureus infected groups...... 187

Table 4-5: The discriminatory polypeptides identified from CE-MS analysis that were significant in the basis of comparison between control and infected groups...... 189 10

Table 4-6: The discriminatory polypeptides identified from CE-MS analysis that were significant in the basis of comparison between two infected groups (E. coli and S. aureus)...... 190

Table 4-7: The discriminatory polypeptides identified from CE-MS analysis that were significant in the basis of both comparison (between control and infected groups as well as between E. coli and S. aureus)...... 191

Table 4-8: The amino acid sequences of potential biomarkers from the LC-MS/MS analysis...... 195

Table 4-9: The protein identities obtained from the UniProtKB database...... 198

Table 5-1: The acute phase proteins (SAA and Hp) concentrations and SCC measurement in all control milk samples used in this study...... 216

Table 5-2a: The acute phase proteins (SAA and Hp) concentrations and bacterial causing mastitis in infected milk samples caused by E. coli used in this study...... 217

Table 5-2b: The acute phase proteins (SAA and Hp) concentrations and bacterial causing mastitis in infected milk samples caused by S. aureus used in this study...... 218

Table 5-4: The first 20 of metabolites with the highest m/z intensity of positive ionisation mode from the total of 814 metabolites identified using the MzMine software in control, E.coli infected and S. aureus infected groups...... 223

Table 5-5: The first 20 of metabolites with the highest m/z intensity of negative ionisation mode from the total of 814 metabolites identified using the MzMine software in control, E.coli infected and S. aureus infected groups...... 225

Table 5-6: The top 20 metabolites by m/z intensity of positive ionisation mode analysis in control group...... 226

Table 5-7: The top 20 metabolites by m/z intensity of positive ionisation mode analysis in E. coli group...... 227

Table 5-8: The top 20 metabolites by m/z intensity of positive ionisation mode analysis in S. aureus group...... 228

Table 5-9: The top 20 metabolites by m/z intensity of negative ionisation mode analysis in control group...... 229

Table 5-10: The top 20 metabolites by m/z intensity of negative ionisation mode analysis in E. coli group...... 230

Table 5-11: The top 20 metabolites by m/z intensity of negative ionisation mode analysis in S. aureus group...... 231 11

Table 5-12: Metabolites identified from positive ionisation mode analysis which were absent in control group but found greater in S. aureus infected group by means of m/z intensity than E. coli infected group...... 232

Table 5-13: Metabolites identified from positive ionisation mode analysis which were absent in control group but found greater in E. coli infected group by means of m/z intensity than S. aureus infected group...... 233

Table 5-14: Metabolites identified from negative ionisation mode analysis which were absent in control group but found greater in S. aureus infected group by means of m/z intensity than E. coli infected group...... 235

Table 5-15: Metabolites identified from negative ionisation mode analysis which were absent in control group but found greater in E. coli infected group by means of m/z intensity than S. aureus infected group...... 236

Table 5-16: A relative change of metabolites in positive ionisation mode analysis in E. coli infected group in comparison to control group...... 238

Table 5-17: A relative change of metabolites in positive ionisation mode analysis in S. aureus infected group in comparison to control group...... 240

Table 5-18: A relative change of metabolites in negative ionisation mode analysis in E. coli infected group in comparison to control group...... 242

Table 5-19: A relative change of metabolites in negative ionisation mode analysis in S. aureus infected group in comparison to control group...... 244

Table 5-20: A relative change of metabolites in positive ionisation mode analysis in E. coli and S. aureus infected groups in comparison to control group...... 246

Table 5-21: A relative change of metabolites in negative ionisation mode analysis in E. coli and S. aureus infected groups in comparison to control group...... 249

Table 5-22: Six metabolic pathways in lipid metabolism with the most apparent changes of metabolites from both positive and negative ionisation mode analyses...... 258

Table 5-23: Six metabolic pathways in carbohydrate metabolism with the most apparent changes of metabolites from both positive and negative ionisation mode analyses...... 259

Table 5-24: Six metabolic pathways in protein metabolism with the most apparent changes of metabolites from both positive and negative ionisation mode analyses...... 260

Table 5-25: Examples of di- and tri-peptides which were found increased in both mastitic groups from both positive and negative ionisation mode analyses...... 261

List of Figures

Figure 2-1: Immuno dot blot of biotinylated anti-bovine Hp reacting with bovine Hp at dilutions of 1:100, 1:200, 1:400, 1:800 and 1:1600...... 107

Figure 2-2: Optimisation of coating antibodies concentrations using different concentrations for the developed ELISA...... 108

Figure 2-3: Optimisation of biotinylated antibody concentrations for the developed ELISA...... 109

Figure 2-4: Standard curve obtained from the optimised first and second batch of biotinylated antibodies (3.43 mg/ml) diluted at 1:1000...... 110

Figure 2-5: Standard curve obtained from optimised standard bovine Hp dilution...... 112

Figure 2-6: Linearity of 3 bovine milk samples after serial dilution...... 114

Figure 2-7: Ouchterlony immunodiffusion gel of anti-bovine Hp...... 118

Figure 2-8: The box-plot showed the distribution of SAA levels in bovine milk samples between healthy, subclinical mastitis and clinical mastitis groups...... 121

Figure 2-9: The box-plot showed the distribution of Hp levels in bovine milk samples between healthy, subclinical mastitis and clinical mastitis groups...... 122

Figure 2-10: The box-plot showed the distribution of SCC values in bovine milk samples between healthy and subclinical mastitis groups...... 123

Figure 3-1: 1D gel electrophoresis whey samples after various methods of casein precipitation in comparison with whole milk sample without any treatment...... 149

Figure 3-2a: 1D gel electrophoresis of 5 individual subclinical mastitic and 6 individual healthy milk samples...... 151

Figure 3-2b: 1D gel electrophoresis of 1 subclinical mastitic and 1 healthy milk samples...... 152

Figure 3-2c: 1D gel electrophoresis of 1 subclinical mastitis and 1 healthy milk samples...... 153

Figure 3-3a: 2D gel electrophoresis of pooled subclinical mastitic whey sample...... 155

Figure 3-3b: 2D gel electrophoresis of pooled healthy whey sample...... 156

Figure 3-4a: Difference gel electrophoresis on clinical mastitic and healthy milk samples...... 158

Figure 3-4b: Difference gel electrophoresis on pooled clinical mastitic milk sample on Cye5 fluorescent dye...... 159 13

Figure 3-4c: Difference gel electrophoresis on pooled healthy milk samples on Cye3 fluorescent dye...... 160

Figure 4-1: The polypeptide fingerprints of both non-infected and infected groups...... 182

Figure 4-2: The polypeptide fingerprints of both E. coli and S. aureus groups...... 182

Figure 4-3: The model of polypeptide fingerprint after multiple testing of non-infected and infected groups...... 184

Figure 4-4: The first model of polypeptide fingerprint after multiple testing of E. coli and S. aureus groups...... 184

Figure 4-5: The second model of polypeptide fingerprint after multiple testing of E. coli and S. aureus groups...... 185

Figure 4-6: The third model of polypeptide fingerprint after multiple testing of E. coli and S. aureus groups...... 185

Figure 4-7: Milk polypeptides detected in CE-MS analysis based on the number of basic amino acids found in the amino acid sequence...... 197

Figure 5-1: An example of various metabolic pathways generated by PATHOS website...... 254

Figure 5-2: An example of metabolic pathway (Arachidonic acid metabolism) which listed down the metabolites with greatest changes of abundances (dark red colour) to the least changes of abundances in both groups of mastitis (light blue colour; not showing)...... 255

Figure 5-3: An example of histogram of prostaglandin F2 alpha showing abundances in control, E. coli infected and S. aureus infected groups...... 256

Figure 5-4: An example of KEGG map generated from arachidonic acid metabolism to highlight the potential metabolites involved in a particular metabolic pathway...... 257

List of Abbreviations

% percentage ~ approximately £ Pound sterling + plus < less than > more than °C degree Celcius

15d-PGJ2 15-Deoxy-Delta-12,14-prostaglandin J2 2D two dimensional 2DE two dimensional electrophoresis 2D-PAGE two dimensional polyacrylamide gel electrophoresis 4CN 4-chloro-1-naphthol AA amyloid A ABP activity-based probe ABS antibody binding solution ACE angiotensin converting enzyme ACN acetonitrile AcP acid phosphatase ADP adenosine diphosphate AGP acid glycoprotein Ala alanine AMP adenosine monophosphate amu atomic mass unit AP alkaline phosphatase APP acute phase protein APR acute phase reaction/response AQUA absolute quantification Arg arginine 15

ASAT aspartate aminotransferase ASG acid-soluble glycoprotein Asp aspartic acid ATP adenosine triphosphate ATPase adenosine triphosphatase AUC area under curve BCA bicinchoninic acid bcSCC Box-Cox transformed somatic cell count BH Benjamini-Hochberg BHB beta-OH butyrate BHBA beta hydroxybutyrate bHp bovine haptoglobin BMC buttermilk concentrate BMSCC bulk milk somatic cell counts BSA bovine serum albumin C complement CATHL1 cathelicidin 1 CD cluster of differentiation CE capillary electrophoresis cells/ml cells per millilitre CE-MS capillary electrophoresis mass spectrometry CFB complement factor B CHAPS 3-[(3-cholamidopropyl)dimethylammonio] propanesulfonate CHCA cyano-4-hydroxy cinnamic acid Chem chemical cIEF capillary isoelectric focusing Cl- chloride ion CL-43 collectin-43 CLU Clusterin CM clinical mastitis 16

CMT California Mastitis Test CN Casein co. corporation ConA- Concanavalin A COX-2 cyclooxygenase-2 CRP C-reactive protein CPLL combinatorial peptide ligand library Cu copper CV Coefficient of variance Cy cyanine CyDye cyanine dye Cys cysteine Da Dalton DiGE difference gel electrophoresis DIM days in milk Dm error difference between the theoretical mass value of the peptide and the experimental mass DNA deoxyribonucleic acid DTT dithiothreitol E.coli Escherichia coli e.g example EB Equilibration buffer EC electrical conductivity ELISA enzyme-linked immunosorbent assay ESI electrospray ionization ESI-MS electrospray ionization mass spectrometry EU European Union FA fatty acid Fe iron FFFA free fatty acid FGF fibroblast growth factor 17

FGFBP fibroblast growth factor binding protein FTICR fourier transform ion-cyclotron resonance FTMS Fourier transform mass analyzer g/litre gram per litre GC-MS gas chromatography mass spectrometry Gly glycine GlyCAM-1 glycosylation-dependent cell adhesion molecule 1 Glu glutamic acid GSH glutathione GSHPx glutathione peroxidase H histidine

H2O2 hydrogen peroxide Hb haemoglobin HbCN cyanmethaemoglobin HDL high density lipoprotein Hb-Hp haemoglobin-haptoglobin HCl acid hydrochloride His histidine HMW high molecular weight HNMR proton nuclear magnetic resonance Hp haptoglobin Hp-HbCN haptoglobin-cyanmethaemoglobin HPLC high-performance liquid chromatography Hp-MMP9 haptoglobin-matrix metalloproteinase HSA human serum albumin ICAT Isotope-coded affinity tag ID identification IEF isoelectric focusing IEX ion exchange chromatography Ig immunoglobulin 18

Ile isoleucine IL-1 interleukin-1 IL-10 interleukin-10 IL-12 interleukin-12 IL-19 interleukin-19 IL-1β interleukin-1 beta IL-6 interleukin-6 IL-8 interleukin-8 IMI intramammary infection Inc. incorporation iNOS inducible nitric oxide synthase IPG immobilized pH gradient IQR inter-quarter ratio IRT infrared thermography iTRAQ isobaric tags for relative and absolute quantitation K lysine K+ potassium ion kDa kilo Dalton KEGG Kyoto Encyclopaedia of Genes and Genomes kV kilo volt LAB lactic acid bacteria lb pound LBP lipopolysaccharide binding protein LC liquid chromatography LC MS liquid chromatography mass spectrometry LC-HSRM-MS liquid chromatography-high-resolution selected reaction monitoring- mass spectrometry LC-MRM-MS/MS liquid chromatography-multiple reaction monitoring-tandem mass spectrometry LC-MS/MS liquid chromatography tandem mass spectrometry LDH lactate dehydrogenase 19

Leu leucine Lf lactoferrin

LKA4 leukotriene A4

LKB4 leukotriene B4 LMW low molecular weight logSCC logarithmic somatic cell count logNAGase log N-acetyl-beta-D-glucosaminidase Lp lactoperoxidase L-PGDS lipocalin-type prostaglandin D synthase LPL lipoprotein lipase LPS lipopolysaccharide LTA lipotechoic acid

LTB4 Leukotriene B4 Lys lysine m or M methionine M molar MALDI matrix-assisted laser desorption ionization MALDI TOF MS matrix-assisted laser desorption ionization-time of flight mass spectrometry MALDI-TOF matrix-assisted laser desorption ionization-time of flight tandem mass MS/MS spectrometry Met methionine MFGE8 milk fat globule EGF factor 8 MFGM milk fat globule membrane mg/ml milligram per millilitre MH+ protonated mass of peptide MHC major histocompatibility complex MHp milk haptoglobin min minute ml millilitre mM millimolar MPO myeloperoxidase 20

Mr relative molecular mass mRNA messenger ribonucleic acid MS mass spectrometry MS mass spectrometry mS/cm milliSiemens per centimetre MS/MS tandem mass spectrometry M-SAA3 mammary-associated serum amyloid A3 Mud-PIT multidimensional protein identification MUFA monounsaturated fatty acid MWCO molecular weight cut off m/z mass to charge ratio n total Na+ sodium ion NaCl sodium chloride NAD nicotinamide adenine dinucleotide NADH reduced nicotinamide adenine dinucleotide NAGase N-acetyl-β-D-glucosaminidase

NaHCO3 sodium bicarbonate NaOH sodium hydroxide NCBI National Centre for Biotechnology Information NEB negative energy balance NEFA non-esterified fatty acid nESI-MS/MS nanoelectrospray-tandem mass spectrometry NET neutrophil killing mechanism ng/ml nanogram per millilitre

NH4 ammonium

NH4OH ammonium hydroxide

NH4SO4 ammonium sulphate NHS N-hydroxysuccinimidyl NHS-PEO4 N-Hydroxysuccinimide polyethylene oxide 4 21

NK-κB nuclear factor kappa-light-chain-enhancer of activated B cells nm nanometre NMR nuclear magnetic resonance NO nitric oxide NOS nitric oxide synthase NPN non-protein nitrogen

- O2 superoxide anion OD optical density OH hydroxyl radical P or p proline P4 progesterone PAGE polyacrylamide gel electrophoresis para-κ-CN para-kappa casein PBS phosphate buffered saline PCM milk prostacyclin PCR polymerase chain reaction

PGD2 prostaglandin D2

PGE2 Prostaglandin E2 PGF1α prostaglandin F1 alpha PGF2α prostaglandin F2 alpha

PGI2 prostacyclin

PGJ2 prostaglandin J2 PGLYRP1 peptidoglycan recognition protein 1 pH power of hydrogen PHA- phytohaemagglutinin pI isoelectric point Pm probability of misclassifying PMF peptide mass fingerprinting PMN polymorphonuclear PMNL polymorphonuclear leukocyte 22 pmol/μl picomol per microlitre ppm parts per million PPV positive predictive value Pro proline PSD post-source decay psi pounds per square inch PTM post transitional modification PUFA polyunsaturated fatty acid p-value probability that null hypothesis is true Px peroxidase Q quadrupole Q-TOF MS/MS quadruopole-time of flight tandem mass spectrometry r correlation R arginine RAM restricted access media RP reversed phase RP-HPLC reversed phase-high performance liquid chromatography RSNO S-nitrosothiols RT rectal temperature RT-PCR reversed transcription polymerase chain reaction RBP retinol binding protein S. agalactiae Staphylococcus agalactiae S. aurues Staphylococcus aureus S. epidermidis Staphylococcus epidermidis S. uberis Streptococcus uberis S. dysgalactiae Streptococcus dysgalactiae SAA serum amyloid A SCC somatic cell counts SCX strong cation resin SD Standard deviation 23

SDPP short-day photo period SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis Se selenium SELDI-TOF Surface-enhanced laser desorption ionization-time of flight Ser serine SFA saturated fatty acid SHp serum haptoglobin SILAC Stable isotope labelling with amino acids in cell culture SNF solid non-fat SOD Superoxide dismutase spp species SPE solid phase extraction SPME solid phase microextraction SPR surface plasmon resonance SRID single radial immunodiffusion SST skin surface temperature SVM support vector machine T trace TBS Tris-buffered saline TCA citric acid cycle TFA trifluoroacetic acid TLR toll-like receptor TMB Tetra methyl benzidine TNF-α tumour necrosis factor-alpha TOF time-of-flight Tris-HCl Tris-hydrochloride TrpRS tryptophanyl-tRNA-synthetase TTR transthyretin

TXA2 thromboxane A2

TXB2 thromboxane B2 24

UFA unsaturated fatty acid UK United Kingdom UPLC ultra-high pressure liquid chromatography USA United States of America utlIEF ultra-thin layer isoelectric focusing UV ultra violet UV/VIS ultra violet visible V voltage Val valine vs versus v/v volume over volume w/v weight over volume WPC whey protein concentrate WST Whiteside test X times XCorr cross correlation value from the search x g centrifugal force in gravity Zn zinc z-value standard score α1-AGP alpha 1 acid glycoprotein

α1-PI alpha 1-acid proteinase α-CN alpha casein α-LG alpha lactoglobulin αS1-CN alpha S1 casein αS2-CN alpha S2 casein β-CN beta casein β-Gase β-glucuronidase β-LA beta lactalbumin γ gamma

γ1-CN gamma 1 casein 25

γ2-CN gamma 2 casein

γ3-CN gamma 3 casein

δ-12-PGJ2 delta-12-prostaglandin J2 δ-CN delta casein κ-CN kappa casein μg microgram μg/ml microgram per litre μl microlitre

Acknowledgements

In the name of Allah SWT, the Most Gracious and the Most Merciful, I offer my humble thank gratitude You for giving me the strength on completing this PhD thesis.

This thesis would not have been possible without the guidance and the help of several individuals who in one way or another contributed and extended their valuable assistance in the preparation and completion of this study.

My PhD study would not have been started with the sponsorship from the Ministry of Higher Education Malaysia and Universiti Putra Malaysia for their support in terms of living allowances in the UK. My colleagues and staff in Faculty of Veterinary Medicine, Universiti Putra Malaysia, thank you for your constant moral support throughout my period of study.

I offer my utmost gratitude to Professor David Eckersall, my principal supervisor who has supported me throughout my thesis with his patience and knowledge whilst allowing me the room to work in my own way. I attribute the level of my PhD degree to his encouragement and effort and without him this thesis, too, would not have been completed or written. One simply could not wish for a better or friendlier supervisor. Alongside David, I thank my co-supervisor Professor David Barrett without whose broad knowledge in the veterinary field and assistance, this thesis would not have been successful.

I am also extremely grateful to the dairymen and staff involved at Cochno Farm and Vale Veterinary Laboratory, UK for providing me plenty of milk samples throughout my study. Without their help, my work would not have been completed.

Special thanks to Rahul Nelli for offering me his help during my first few weeks of my study with several of laboratory works. I also wish to thank Dr. Tim Parkin for his expertise in statistical analyses.

In the various laboratories, I have been aided by many important people and I wish to express gratitude to Dr. Laura Bence and Mary Waterston for their help especially with the milk acute phase protein assays. Dr. Richard Burchmore, Kamburapola Jayawardena and Michael Walsh for their help and knowledge in my proteomic works, Dr Karl Burgess who had helped me in my metabolomic work and lastly to Dr William Mullen and Dr Amaya Albalat for their continuous help and support in my peptidomic work. 27

My special thanks to my husband, Mr Mohd Hasan Mohd Saaid for his constant help, understanding, support and encouragement. Without you, I could never have completed this study. And to our daughter, Nur Adaira Safiyyah thank you for giving me a constant inspiration and happiness while writing up this thesis, I am also indebted to my family (parents and sister) whose encouragement was never end.

Finally, I extend my apologies to everyone whom I have neglected in the name of my PhD. Many thanks, to my fellow Malaysian for their support and help. I am truly grateful for every chance and opportunities that have been given in my life. I look forward to more exciting life beyond this PhD.

Rozaihan Mansor, February 2012 28

Author’s Declaration

The work presented in this thesis was performed solely by the author except where the assistance of others has been acknowledged.

Rozaihan Mansor, February 2012

Dedication

To my daughter, Nur Adaira Safiyyah Mohd Hasan

Chapter 1, 30

Chapter I. Introduction

1.1 Bovine Mastitis and its Impact on the Dairy Industry

Mastitis is an inflammation of mammary gland parenchyma which is characterized by a range of physical and chemical changes of the milk and pathological changes in the udder tissues (Radostits et al. 2000). Significant milk changes that can be observed in bovine mastitis are the presence of clots in milk, milk discolouration and high numbers of leukocytes in affected milk. Furthermore, apparent clinical signs in bovine mastitis comprise swelling, heat and pain in the udder.

Mastitis is usually caused by bacterial pathogens which can be classified into two groups; the contagious pathogens which include Streptococcus agalactiae, Staphylococcus aureus and Mycoplasma bovis as well as environmental pathogens which include Streptococcus species (Streptococcus uberis and Streptococcus dysgalactiae) and environmental coliforms (Gram negative bacteria Escherichia coli, Klebsiella spp., Citrobacter spp., Enterobacter spp., Enterobacter faecalis and Enterobacter faecium., and other gram negative bacteria such as Serratia, Pseudomonas and Proteus (Radostits et al. 2000)

The bovine mammary gland is protected by innate and specific immune responses (Sordillo & Streicher 2002). However, factors such as environmental and physiological could compromise the defence mechanism of the mammary gland (Hopster et al. 1998; Waller 2000). Milking by using machines can also contribute to teat damage which in turn increases the susceptibility of mastitis-causing pathogen colonization. Poor housing environment, increased cow densities per unit and low ventilation can also increase the susceptibility to bovine mastitis. Nevertheless, the most important factor in contributing to increase bovine mastitis susceptibility is the lactation stage of a cow. It is known that during the periparturient period, the immunity of the udder is compromised due to physiological changes that happen in preparation for lactation (Oliver & Sordillo 1988).

Mastitis is one of the most prevalent diseases in dairy cows. It is an endemic disease and happens to be the most frequent and most costly disease affecting dairy herds worldwide (Halasa et al. 2007; Miller et al. 1993). It affects the quality of the milk through changes in milk composition and in return, affects the economy of dairy industry. The United Kingdom (UK) is the third largest milk producer in the EU after Germany and France and Chapter 1, 31 ninth largest producer in the world (Defra, Agriculture in the United Kingdom 2009, April 2010, Table 9.1). Around one million cases of bovine mastitis occur each year in the UK with £200 million of losses in production and treatments every year (Science Daily 2008).

Various papers have discussed the economic impact of bovine mastitis. To simplify their conclusions, economic losses due to mastitis can be caused by milk production losses, use of treatments, discarded milk, veterinary services, labour, product quality, materials and investments, diagnostics, other diseases and culling (Halasa et al. 2007). Persistent decrease in milk production is the main detrimental effect that contributes to the economic impact of mastitis (Seegers et al. 2003). However, it is quite difficult to estimate the economic losses caused by bovine mastitis as they are influenced by various factors.

According to Bhikane & Kawitkar (2000), this most important disease in dairy cattle can cause up to 70% of reduced milk production, 9% of milk discard after treatment, 7% of the cost of veterinary services and 14% of premature culling.

Traditionally, the control of mastitis has involved the use of chemical disinfectants, antiseptic or herbal teat dips (Maiti et al. 2004) and antibiotic therapy. Antibiotic therapy has been used in mastitis control for about 50 years (Sharma 2007). However, it was suggested that antibiotic therapy was not actually helping to reduce the incidence of mastitis. In fact, several problems arise from the use of antibiotics for example, developing resistance to antibiotic, questionable drug efficacy and presence of antibiotic residues in the milk. Thus, Sharma (2003) suggested the use of alternative approaches to control mastitis that enhance the immunity of mammary gland by using immunoregulatory micronutrients (Vitamin E and selenium), vaccines and cytokines.

1.1.1 Bovine Milk Milk is one of the important biological fluids and contains considerable amounts of nutrients. All types of milk contain proteins in colloidal dispersion as micelles, fats in emulsified globules coated with a membrane and easily digested, minerals, vitamins and other components (Jensen et al. 1991, 1995). Each of these components plays beneficial roles to the host.

Colostrum is another type of milk that is produced by cows during the first several days postpartum. It is important for the growth and health of the neonates as it contains antibody (immunoglobulins) and complement that provide antimicrobial properties against different types of pathogens which can be detrimental to the newborn especially during early age Chapter 1, 32

(Korhonen et al. 2000). Thus, colostrum provides nutrients to the newborn as well as conferring passive immunity to them. However, in this thesis, we will focus only on milk protein of bovine.

1.1.2 Milk Protein Proteins are arguably the most important nutritional factors in the milk. In bovine, six major gene products of the mammary epithelial cells have been identified which include

αS1-, αS2-, β-, κ-caseins, β-lactoglobulin and α-lactalbumin (Swaisgood 1982, 1993). Apart from the presence of blood plasmin and plasminogen in the milk, small peptides (e.g; proteose-peptone which is largely derive from the N-terminus of β-casein and from fat globule membrane glycoproteins) and large peptides (e.g; γ-caseins are derived from the C-terminus of β-casein) are also present in milk with varying concentrations (Swaisgood 1993) as well as a number of low abundance minor proteins and enzymes (plasmin and lipoprotein lipase).

Protein is one of the milk compositions known to affect more of the properties of milk and most dairy products than any other constituent, thus considerable amounts of research have been performed to investigate their unique properties and technological importance.

Milk contains acid-precipitated proteins known as casein (CN) which are converted to curds by rennin (Kinsella et al. 1984). The liquid left after casein precipitation of skim or whole milk is called whey proteins. Another two groups of proteinaceous materials are proteose peptones and non-protein nitrogen (NPN) which was discovered by Rowland (1938).

Pedersen and Mellander (McMeekin 1970) successfully resolved isoelectric casein into three proteins namely α-, β-, γ- caseins depending on their electrophoretic mobility. Alpha- casein was further resolved by Waugh and van Hippel (1956) into αs-casein which contains two types of proteins αs1- and αs2- caseins as well as κ-casein. According to Fox &

Brodkorb (2008), α s1-casein, α s2-casein, β-casein and κ -casein represent approximately 38, 10, 35 and 12% respectively of whole bovine casein.

As for the whey proteins, they are globular molecules in which the acidic/basic and hydrophobic/hydrophilic amino acids are evenly distributed along their polypeptide chains (Evans 1982). Whey proteins comprise of β-lactoglobulin (β-LG), α-lactalbumin (α-LA), immunoglobulins (Igs), lactoferrin and lactoperoxidase, together with other minor components (Madureira et al. 2007). Chapter 1, 33

Other proteins that are present in bovine milk comprise metal-binding proteins (glutathione peroxidase), β-microglobulin, osteopontin and vitamin-binding proteins, angiogenins, kininogen, glycoproteins, proteins in the milk fat globule membrane and growth factors.

Milk Proteins Examples

Caseins αS1-casein, αS2-casein, β-casein, κ-casein Whey proteins β-lactoglobulin, α-lactalbumin, immunoglobulins, lactoperoxidase Others metal-binding proteins, β-microglobulin, osteopontin, vitamin-binding proteins, angiogenins, kininogen, glycoproteins, proteins of milk fat globule membrane and growth factors

Table 1-1: Examples of caseins, whey and other proteins present in bovine milk composition

1.2 Diagnostic Methods for Bovine Mastitis These traditional diagnostic methods discussed below are the most common ways of diagnosing bovine mastitis at the farm level worldwide. They are known to be used widely in detecting bovine mastitis. Despite of several major drawbacks in these methods, they still are widely used in diagnosing bovine mastitis in dairy herds.

1.2.1 Somatic Cell Counts Somatic cell count (SCC) has proved to be one of the most useful diagnostic techniques to detect the presence and occurrence of bovine mastitis and especially for bovine subclinical mastitis (Dohoo & Meek 1982; Schukken et al. 2003). According to Shook (1989), currently, the dairy industry relies on this method for milk quality payment schemes (Viguier et al. 2009) where elevated levels of somatic cells result in low payments. Automated devices have been developed recently so that rapid determination of somatic cell counts can be performed by using the Coulter Milk Cell Counter; where the particles are counted through an electric field and the Fossomatic; where the cells are stained using fluorescent dyes before counting the fluorescent particles (Dohoo & Meek 1982).

Normally, in a non-infected mammary gland, the SCC is low and macrophages are the predominant cells in the milk, which account for 60% of the total somatic cells with 12% Chapter 1, 34 consisting of neutrophils and 28% of lymphocytes (Burvenich et al. 1995; Napel et al. 2009). However, following microbial infection, approximately within 6 hours after the release of bacterial endotoxin, the somatic cells increase dramatically from fewer than 105 cells/ml of milk to an excess of 106 cells/ml (Kehrli and Shuster 1994). At this stage, the majority of cell populations are neutrophils (>90%), which remain the main cell type up to 59 hours following infusion (Saad & Ostensson 1990). However, according to Van Werven et al. (1997), it may take about 3 weeks for the SCC to return to its normal concentration following successful bacterial elimination. Thus, the levels of neutrophils as the percentage of the SCC can be used as a mastitis indicator (Hamann and Kromker 1997). Leukocytes, primarily neutrophils are recruited from the blood in order to fight against bacterial colonization and thus making neutrophils the cells predominant in milk somatic cells during mastitis (Leitner et al. 2000).

The presence of increased somatic cell counts can be explained by the pathogenesis of bovine mastitis. Following microbial establishment in the mammary gland, the bacteria will stimulate both the cellular and humoral defence mechanisms of the udder and will further deteriorate its condition by multiplying and liberating toxins. Endotoxin released by the bacteria induces leukocytes and epithelial cells to release chemoattractants such as cytokines, interleukins (IL-1, IL-8), eicosanoids (PGF2α), oxygen radicals and acute phase proteins (APPs). Polymorphonuclear cells (PMNs) are attracted to the site of inflammation to engulf and destroy the invading bacteria and in turn, release intracellular granules which contain bactericidal peptides, proteins, enzymes, and neutral and acidic proteases. The apoptotic PMNs are later engulfed and ingested by macrophages and the dead and sloughed off mammary epithelial cells together with dead leukocytes give rise to high somatic cell counts in the milk

To correctly interpret the SCC, several factors need to be considered. First, is the infection status of the quarter of the mammary gland. The SCC depends on the causal organisms (commensal organisms; Corynebacterium bovis or coagulase negative streptococci or major pathogens; Streptococci species, Staphylococcus aureus and coliforms) as different types of pathogens, result in different ranges of SCC. Second, the number of quarters infected is particularly important in composite (cow) SCC as the high cell count milk from infected quarters is diluted with low cell count milk from uninfected quarters. Increased age of a cow has been reported to be associated with an increase of cellular content of milk due to increased prevalence of disease in older cows (Marshall & Edmondson 1962; Chapter 1, 35

Reichmuth 1975; Dohoo & Meek 1982). After calving, there is a possibility of the somatic cell counts rising in uninfected quarters and then decreasing to the normal level 4-5 days postpartum (Dohoo & Meek 1982; Barkema et al. 1999). Some authors have reported that the somatic cell counts increase as the lactation progresses (Schultz 1977). Towards the end of lactation, SCC is seen to be elevated (Brolund 1985; Miller & Paape 1988) but return to a low level if the quarter is uninfected. However, if there is a presence of infection, the SCC remains high and this can be used to detect a new intramammary infection postpartum (Barkema 1999). Thus, stage of lactation seems to affect the interpretation of SCC.

Milking frequency also seems to affect the SCC. Increasing the frequency of milking (from two times a day to three times a day) decreased the SCC of the bulk milk (Hogeveen et al. 2001). Cell counts were seen to be elevated in the afternoon rather than in the morning (Kukovics 1996) and also they appeared to be highest in the milk strippings. These cell count concentrations could persist for up to 4 hours before return to low levels (Smith & Schultze 1967). Other factors that should be considered include season, stress, day to day variation, technical aspects and management (Dohoo & Meek 1982).

The type of sample taken for SCC analysis (quarter, composite or bulk tank) is an important factor when interpreting the results. One should remember that there are possibilities of having false-positive or false-negative results. Infected cows sometimes have low cell counts and vice versa. Second, cell counts are only an indication of the presence of infection within the mammary gland, rather than indicating the presence of pathogen. Finally, the evaluation of herd averages is better than the evaluation of individual quarter, cow or bulk tank level (Dohoo & Meek 1982).

With quarter milk samples, the threshold of 200 000 cells/ml has been recommended to classify the quarter as being infected (with a major pathogen) (Hillerton 1999a). When interpreting the composite SCC, the dilution effect of milk from uninfected quarters on the elevated counts from infected quarters should be taken into consideration. It has been suggested that the logarithmic transformation of the SCC can be done before the average of cell count is calculated (Dijkman 1975). Bulk tank SCCs seem to be a good indicator of the general state of udder health and levels below 250 000 cells/ml indicate a good state of udder health and over 500 000 cells/ml indicate the presence of severe mastitis in a herd (Dohoo & Meek 1982). Chapter 1, 36

High somatic cell counts affect the milk quality and milk production as well as milk composition. Due to high SCC, the quality of milk is altered as it causes deterioration in the flavour quality and the shelf life of the milk. These effects result from the breakdown of milk protein and fat and an increase in acid degree value (Ma et al. 2000). High levels of whey protein and low casein levels lead to lower cheese yields (Sharif & Muhammad 2008) because an elevation of rennet to cutting time and reduced curd firmness are found. In turn, severity of mastitis affects the value of food in terms of low protein and fat contents in milk (Ullah et al. 2005). Thus, an increase in SCC is an indicator of losses in milk production due to subclinical mastitis. However, it must be remembered that other factors associated with mastitis could also lead to production losses such as discarded milk, drug costs, veterinary fees, extra labour, increased cow replacement costs and loss of genetic potential (Dijkman 1975). An intramammary infection during mastitis can cause considerable impact on the mammary gland of a cow. Permeability of the blood-milk barrier is disrupted and results in increased leakage of blood components into milk such as serum albumin and immunoglobulins (Haenlein et al. 1973). Furthermore due to infection, the synthetic activity of mammary tissue seems to be affected, thus decreasing lactose fat and α-lactalbumin level in milk (Harmon 1994).

1.2.2 California Mastitis Test (CMT) A rapid, simple test to detect mastitis called the California Mastitis Test (CMT) was developed by Schalm and Noorlander in 1957. In this test, a mixture of the reagent/detergent and milk caused precipitation and gel formation which reflected the cell count of the milk (Barnum & Newbould 1961). According to Carroll and Schalm (1966), deoxyribonucleic acid (DNA) originating from the nuclei of somatic cells constituting the inflammatory exudates cause this CMT reaction. Jensen (1957) reported that this CMT reaction was due to the formation of a gel of leukocyte proteins and the reaction was negative until the level of cell count exceeded 500 000 cells/ml. Thus, CMT is one of the popular cow-side mastitis tests to indicate the presence of mastitis based on the leukocyte count in the infected milk.

Detergents used in the CMT decrease the surface tension, change the structure and conductivity of cell membranes and the nucleus of the cells. Other than that, it disturbs the osmotic balance, blocks oxidizing reactions and stimulates proteolytic enzymes and increases milk viscosity, leading to gel formation (Sargeant et al. 2001; Ruegg & Reinemann 2002; Middleton et al. 2004). According to Barnum & Newbould (1961), the test was performed by pouring the well-mixed milk to a volume of 3ml in one part of a Chapter 1, 37 partitioned plastic paddle before adding approximately 3ml of the commercial CMT reagent. The paddle can hold 4 samples at one time and is rotated quickly by hand 10 times and graded as either negative, trace (T), 1+, 2+ and 3+. The description is as follows:

CMT reactions Description Negative No change in consistency. T-Trace No visible change in consistency, but when paddle is tipped a slime is momentarily seen on the bottom. 1+ A gel or thick slime forms, but when the paddle is swirled the solution does not move into the centre. 2+ A thick lumpy gel forms, which, when swirled, quickly moves toward the centre 3+ A distinct gel forms which tends to adhere to the paddle, and during swirling a distinct central peak forms.

Table 1-2: Description on the changes of milk after adding the CMT reagent and their CMT scores. (Barnum & Newbould 1961)

Although SCC estimation is a reliable tool to be used to screen and identify intramammary infection (IMI) (Sargeant et al. 2001), determining SCC requires sample submission to a laboratory and is not readily accessible to on-farm use. Thus, CMT serves as an important method of diagnosis that is readily accessible to on-farm use and its scores correlate well with the SCC (Hogan et al. 1999). The CMT and Modified Whiteside (Murphy 1942; Murphy et al. 1941) tests are useful diagnostic tests to identify subclinical mastitis and evaluate its severity (Forster et al. 1966). According to Middleton et al. (2004), if CMT is a reliable test to identify subclinical infected mammary quarters, then it could be used in selecting quarters for cessation of lactation. This therapeutic cessation of lactation in infected mammary quarters can help in decreasing individual cow SCC and thereby bulk tank SCC (Middleton et al. 2004; Middleton & Fox 2001).

Because the CMT reaction is an indication of leukocyte counts, various studies have been performed to correlate the relationship between CMT and leukocyte content of milk. Chapter 1, 38

Barnum & Newbould (1961) reported that CMT showed close association with the leukocyte count as 71.7% of samples with positive CMT were from quarters classified as infected mammary quarters. More samples with less than 500 000 cells/ml gave a trace reaction of CMT. Furthermore, only 2.7% of samples showed a false positive result but indicated that the paddle method of CMT needed experience in both the method of testing and interpretation of the reactions to achieve uniformity and accuracy of the results.

The understanding on the effects of milk composition and yield caused by the varying levels of subclinical mastitis is important in dairy industry. Thus a study by Daniel et al. (1966) determined the effects of subclinical mastitis (as determined by CMT on samples of 24-hour milk production) on milk composition and yield. This study indicated that by one unit of difference in CMT value, the average changes were a decrease in lactose concentration by 0.1%, increased protein level by 0.042%, reduced solid non-fat (SNF) levels by 0.046% and most importantly, reduced milk production by 4.9lb per cow.

One study on the relationship between CMT reaction and milk production and composition of milk from opposite quarters were done by Forster et al. (1966). Before CMT became available, the opposite-quarter comparisons of milk production was established by Crossman et al. (1950) based on that subclinical mastitis was actually associated with decreased milk production. Based on comparisons made from 1258 opposite-quarter milkings from 763 cows, average decreases in milk production were seen according to the CMT reactions (on total quarter milk). The biggest changes of milk production were seen at test reaction of 3 with 43.4% decrease in milk production.

The relationship between CMT reaction and bacteriological analyses has also been studied by Wesen et al. (1967) as it seems probable that both the type of pathogens involved and leukocyte numbers should be considered in diagnosing mastitis. It appeared that there was a relationship between CMT and the type of pathogens identified in a particular quarter as there was at least one type of pathogen found in quarters with different type of CMT reactions. The non-pathogenic pathogens were identified more frequently in quarters with low mastitis test reactions (negative, trace and 1) than in quarters with high mastitis test reactions (2 and 3).

Miller & Kearns (1966) studied the effectiveness of CMT to measure the leukocyte content of quarter milk samples and found that the CMT was reliable for the low leukocyte count milk when the CMT score is 0 with the percent of correct was 90.7. They concluded that as the CMT score increased, the leukocyte content can be expected to increase materially and Chapter 1, 39 that CMT was a reliable tool to estimate leukocyte content from a herd of cows under normal management practices.

To determine the effectiveness of CMT in identifying subclinical mastitis, Middleton et al. (2004) found that the sensitivity of the CMT with a trace result as the cut off for a positive result, was low (<0.61) which indicated CMT was not useful as a screening test as only 50% of infected mammary quarters would yield a positive test result. An ideal screening test should have maximal sensitivity to minimize the false-negative results. Differences on specificity and sensitivity between this study and previous studies are due to the characteristics of the individual pathogens and the prevalence of these pathogens in the herds studied. Staphylococcus aureus for example has different pathogenicity from other bacterial infection which lead to variation in the severity of mastitis they induced (Brookbanks 1966).

Dingwell et al. (2003) indicated that early detection of IMI during early lactation may have significant economic benefits. This author agreed on the idea that a screening test with a high sensitivity (no false-negative CMT reactions) would be ideal to discriminate between non-infected and infected quarters. Although the highest calculated sensitivity and positive predictive value (PPV) were on days in milk (DIM) 4 at a cut-off point of >0, it cannot be concluded that DIM 4 is the best day to perform the CMT because samples were not taken from every cow on every day in this study. Nevertheless, CMT can still be a rapid, accurate and economically feasible test for fresh cows to identify specific pathogens.

Reports by Sharma et al. (2010) were in accordance with other previous studies (Reddy et al. 1998; Tanwar et al. 2001; Goswani et al. 2003) that SCC was the most accurate test to diagnose subclinical mastitis followed by the modified CMT and the modified Whiteside Test (WST) based on the sensitivity and specificity values. They suggest that CMT could be useful as a regular mastitis screening test in dairy herd monitoring programmes even by less trained dairymen as supported by Sargeant et al. (2001).

1.2.3 Enzyme levels

One enzyme that is known to be a reliable and sensitive mastitis indicator is N-acetyl-β-D- glucosaminidase (NAGase) (Kitchen 1976) which has been shown to originate from the cytosol of mammary gland secretory cells (Kitchen et al. 1978). It was first reported by Mellor (1968) that bovine NAGase is derived wholly from leucocytes and its concentration might be convenient to indicate the presence of bovine mastitis. During mastitis, an Chapter 1, 40 increase in cytoplasmic material being shed into milk happens as a result of permeability change in the plasma membranes of the secretory cells together with extensive tissue damage (Chandler et al. 1974).

It has been found that the level of NAGase in milk can be used to estimate the SCC in milk using spectrophotometric measurement (Kitchen & Middleton 1976; Kitchen 1976). However, an improved assay procedure using a fluorimetric substrate has been proposed to facilitate automation of the method (Kitchen et al. 1978). NAGase is one of the indigenous enzymes which found was to be increased in milk during inflammation which originates from phagocytes (Pyorala 2003).

According to Kitchen et al. (1978), the fluorimetric procedure used to analyse NAGase is superior to the previously spectrophotometric assay due to greater sample throughput and ability to get an absolute measure of the level of the product released without problem of sample turbidity problem. Besides, this procedure is suitable for an automated method. They also found good correlation coefficient value between the SCC of 243 quarter foremilk samples determined by the Fossomatic instrument and the fluorimetric assay (r=0.86). This study also concluded that the level of NAGase in milk is significantly influenced by the integrity of the mammary gland secretory cells as the level of NAGase from the mammary gland tissue was the highest compared with other sources listed. Thus, NAGase level in milk could serve as a sensitive indicator of pathological and physiological changes in the mammary gland.

Fitz-Gerald et al. (1981) reported that both NAGase and SCC increased markedly in the milk of first milking after infusion of endotoxin (9-hour interval) in line with a similar increase in the blood component blood serum albumin (BSA) and in Na+/K+. Furthermore, NAGase levels took longer to return to normal than did SCC and BSA and this can be explained due to a healing of injured tissue is a slower process compared with the reversal of other changes.

However, Obara and Komatsu (1984) found that NAGase activity in milk is a better indicator to grade the severity of inflammation than estimating the level of SCC based on the higher correlation between NAGase and chloride (r=.88) and lactose (r = -.80) than correlation between NAGase and SCC (r=.72). In the healthy udders, low levels of NAGase are secreted from the cytosol of secretory cells but during IMI, NAGase is Chapter 1, 41 released from the cells into milk according to intensity of the inflammation. Thus, NAGase provides a useful indicator of the severity of udder inflammation and the injury implicated on the mammary gland. They also found close correlation (r=.91) between NAGase and lactoferrin which is also known to increase in milk from injured epithelial cells after initiation of udder inflammation.

Although the range of NAGase activities were found to be low in bulk milk samples compared with the composite and quarter milk samples (Kitchen et al. 1984a), there was a significant variation of NAGase activities in the bulk milk samples. The NAGase test is more versatile than other alternative mastitis diagnostic methods (BSA, conductivity) as it can be applied to all levels of milk sampling (quarter, composite and bulk), it is rapid and has a simple procedure. The combined use of SCC and NAGase provides a promising and meaningful interpretation of mastitis diagnosis so that herds could be classify according to their level and extent of infection (Pyörälä & Pyörälä 1997).

Previous studies have observed the significance relationship between NAGase and its reference test SCC to assess the inflammatory responses in the mammary gland. However further research work by Kitchen et al. (1984b) has observed the relationship between NAGase and the presence of isolated pathogenic bacteria causing mastitis as the growth of bacterial pathogen in the udder caused secretory disturbances as a results from tissue damage (Schalm et al. 1971; Giesecki 1975). The presence of udder pathogens during mastitis was seen to have a definite relationship with NAGase activities in the infected quarters (Kitchen et al. 1984b). Udders that were infected with minor pathogens had only a slight increased in NAGase activity compared to major pathogens causing mastitis which caused marked increased in NAGase activity. Furthermore, little variation in NAGase activity in the infected milk made it more meaningful in assessing the infection severity than SCC. NAGase levels are also useful in diagnosing bovine subclinical mastitis by applying the composite NAGase analyses on regular monthly basis. However, as previously mentioned, the combination between NAGase and SCC provide a more definitive interpretation on secretory disorders in affected quarters.

Nagahata et al. (1987) have studied the activities of two different enzymes in severe mastitic milk samples NAGase and β-glucuronidase (β-Gase). Beta-glucuronidase was also derived from leukocytes and seems to be a further useful marker in bovine mastitis (Kiermeier & Güll 1966). From their study (Nagahata et al. 1987), they have found Chapter 1, 42 remarkable increases in both NAGase and β-Gase levels in milk samples from acute clinical mastitis cases but only a slight increase in these enzyme levels was observed in milk samples from chronic mastitis. They also assumed that most of NAGase and β-Gase activities found in mastitic milk derived from blood serum, neutrophils and macrophages associated with inflammatory process apart from mammary secretory cells.

Sandholm et al. (1984) proposed the use of milk antitrypsin as a sensitive indicator of bovine mastitis because its activity has shown to be rapidly increased during mastitis. Matilla et al. (1986) indicated that NAGase level was superior to any other tests (BSA, antitrypsin, SCC) when single threshold values were applied. However, when inter-quarter evaluation was applied to avoid background „noises‟, the differentiating abilities of antitrypsin were improved equalling that of NAGase. NAGase also showed highest deflection which is important in monitoring composite and bulk milk samples to minimize the dilution factor in mastitic quarters. The authors concluded that the use of combination between NAGase and antitrypsin to achieve optimal diagnostic potential as mastitis involves in different types of inflammatory reactions (increased leakage of plasma proteins such as antitrypsin and BSA and cell destruction as shown by NAGase).

Teat canal infections can predispose the mammary gland to mastitis as the mammary gland infection is thought to be ascending infections started through the teat canals (Forbes & Herbert 1968). To differentiate between the teat canal infections and mastitic infections, analyses of both cultures of swabs of teat canals and milk samples drawn from lactiferous sinus by teat puncture were done (Murphy & Stuart 1954; Forbes & Hebert 1968; Giesecke & Viljoen 1974). However, Kaartinen & Jensen (1988) claimed that release of NAGase into teat canal secretions could be an indicator of teat canal infections. They found that in S. uberis infected quarter of mammary gland, the NAGase levels in teat sample fractions were significantly lower than in respective milk samples showing that the invasion of S. uberis through teat canal did not cause inflammation or colonization. However, if the quarters were infected with either S. aureus or Micrococcus, NAGase levels in teat sample and in the respective milk samples were at the same level. Thus, these pathogens initially colonize the teat canal before attacking the upper part of milk compartment with the constant bacterial irritation within the streak canal. Analysing the NAGase gradients between teat canals and milk showed a promising evaluation on the inflammatory condition of the teat canal.

Chapter 1, 43

Chagunda et al. (2006) found an increased activity of NAGase in clinically mastitic cows as compared with healthy cows but that other factors should be taken into consideration before interpreting the level of this enzyme. The origins of NAGase have been discussed as some authors claimed that the contribution of NAGase from plasma into milk was not important (Timms & Schultz 1985). However, there has been a report on the increased of NAGase level during dry period in the mammary gland compared with plasma indicating that the enzymes originated from mammary gland (Timms & Schultz 1985).

The presence of PMN leucocytes and macrophages in the mammary gland during subclinical and clinical mastitis induced the phagocytic process with secretion of hydrolytic enzymes and as the consequence, tissue degradation occurred. Among the enzymes found are both nonlysosomals (lactate dehydrogenase and 2-naphthylamilase) and lysosomals (NAGase, β-Gase, β-galactosidase and α-manosidase) (Perdigon et al. 1985). However, β-Gase proved to be the most significant selectively released enzyme in the inflammatory process (Schnyder, & Bagglioni. 1978; Schorlemrner 1977). Studies performed by Perdigon et al. (1985) indicated a high correlation between the SCC and the presence of enzymes β-Gase (76%). This enzyme was also related to the presence of both major pathogens such as S. agalactiae, E. coli, Klebsiella pneumonia, S. aureus as well as minor pathogens (e.g; S. epidermidis). The reaction of β-Gase is very sensitive and because of its high specificity and simplicity, the detection of this enzyme could be applied for mastitis detection.

L-lactate dehydrogenase (LDH) is another type of enzyme that can be used as mastitis indicator. It is involved in the glycolytic pathway and found in the cytoplasm of all cells and tissues in the body (Chagunda et al. 2006). It has 5 isotypes (LDH1 to LDH5) and in milk is known to increase in quantity with mastitis (Bogin et al 1977; Harmon 1994). Chagunda et al. (2006) found that the activity of LDH increased in milk during clinical mastitis compared with healthy cows. However, the enzyme level was influenced by parity, stage of lactation and month of production. Thus, it is important to take these factors into consideration before classifying cows as sick. The association between LDH levels and SCC was stronger in clinically mastitic cows than in healthy cows. It was thought that the origin of elevated LDH was from leukocytes (Bogin 1977) but Kato et al. (1989) found that granulocytes and lymphocytes in mastitic milk might partly contribute to the displacement of LDH isoenzymes into milk. However, Chagunda et al. (2006) indicated that the somatic cells (neutrophils) might contribute to the presence of this Chapter 1, 44 enzyme. When comparing between NAGase and LDH, LDH seem to be better in classifying mastitic cows than NAGase as LDH is a more sensitive indicator in predicting the clinical mastitis.

Another type of enzymes that can be used to detect bovine mastitis is myeloperoxidase (MPO). This enzyme is a lysosomal enzyme which is found in the primary granules of neutrophils (Cooray 1994). Due to its important role in the oxygen-dependent antimicrobial system of neutrophils (Nauseef et al. 1988; Klebanoff, 1991and Cooray et al. 1993), it can be used to estimate neutrophil number as one of the indirect estimator in bovine mastitis. Cooray (1994) found a high correlation between MPO levels in elevated numbers of neutrophils with SCC (r=0.91). Due to its presence in high quantities in mastitic milk, this enzyme has the potential to be an important indicator of bovine mastitis.

1.2.4 Electrical Conductivity

Electrical conductivity (EC) is a measure of the resistance of a particular material to an electric current (Nielen et al. 1992). Normally, milk has a resistance of between 4.0 and 5.5 mS/cm at 22ºC (Wong 1988) and is measured in log normal (Knudsen & Jensen 1990). The concentration of sodium chloride (NaCl) is often expressed as milk electrical conductivity (Kitchen et al. 1980; Linzell & Peaker 1972; Linzell et al. 1974; Peaker 1978).

Conductivity is thus used in diagnosing bovine mastitis, and cow-side tests and on-line measurements have been developed (Nielen et al. 1992). As for on-line measurements, the EC was measured using an electrical cell placed in the milking equipment (Maatje et al. 1983; Puckett et al. 1984). Another system has been developed without electrodes being in contact with the milk but using two toroidal ferromagnetic cores and a milk loop in the system (Onyango et al. 1988).

During mastitis, the destruction of blood capillary permeability, the destruction of tight junctions as well as the failure of the active ion-pumping system causes changes in ion composition in milk. Na+ and Cl- which have high concentrations in extracellular fluid pour into the lumen of mammary gland. In contrast, lactose and K+ leaked into the extracellular fluid and blood (Kitchen 1981; Oshima 1977). These changes caused an increased of EC in mastitic milk (Kitchen 1981). However, several other factors also Chapter 1, 45 influenced the ionic content of milk such as temperature, fat concentration, milk solids and milk fraction (Woolford et al. 1998). These are important factors that should be taken into consideration when interpreting the EC in the diagnosis of bovine mastitis as intramammary infection induces a 15-50% of elevation in EC (Greatrix et al. 1968; Wolfe et al. 1972; Linzell et al. 1974; Fernando et al. 1982).

Electrical conductivity was once assumed to be a difficult method as there was no established value for normal milk making it impossible to interpret the results directly (Little et al. 1968). However, this problem can be overcome if simultaneous comparison was made for all 4 quarters (Greatrix et al. 1968; Wolfe et al.1972) as first proposed by Davis in 1949.

It is thought to be less sensitive than the measurement of sodium (Linzell & Parker 1971, 1972), through daily measurement of foremilk conductivities. Furthermore, it is more practical to automate reading than milk sodium concentrations (Linzell et al. 1974) and it was hoped that daily monitoring of milk conductivity, would allow subclinical mastitis to be detected using this method.

A study by Linzell et al. (1974) indeed indicated that under normal herd management, the use of electrical conductivity can be used as indicator in subclinical mastitis. This study also made it possible to distinguish a value of milk conductivity (56.5mM) above which it referred to severely infected quarters.

Fernando et al. (1982) found that in mastitis caused by major pathogens (S. aureus, coliforms), the conductivity of stripping milk was higher than foremilk conductivity. Furthermore, conductivity of post-milking strippings was more sensitive indicator in intramammary infection than those of foremilks with 3.8% false negative value. In identifying infected and uninfected quarters, a combination of both differential and absolute conductivity provides the most accurate methods. Electrical conductivity (EC) with an in-line system in milking parlour is known to be a rapid and easily obtained measurement.

Further study by Fernando et al. (1985) confirmed that increases of EC, Cl-, Na+, BSA and logSCC were related to intramammary infection. Marked changes were more pronounced in infection caused by major pathogens. It was also confirmed that the EC were higher in Chapter 1, 46 strippings milk than in foremilks. Furthermore, the accuracy of EC was found to be higher in strippings milk than in foremilks. Their results however were in contrast with results of Linzell et al. (1974) as they found accurate results from foremilks. This difference was thought to be due to variation in sampling techniques. Tests for EC, Na+ and Cl- contents were more accurate in detection of infection variables than other predictors. However, their results were in accordance with Woolford & Williamson (1982) and Yamamoto (1982) who found increased sensitivity of strippings to udder infection. Due to increased intramammary pressure from accumulation of milk, there is a possibility of higher concentration of ions in the mastitic milk. Thus it was suggested that EC has the potential as a screening test in bovine mastitis as its accuracy to detect mastitis compared favourably with other indirect mastitis tests (SCC, BSA, Na+, Cl-, etc)

However, EC also showed to be influenced by herd variation. Sheldrake et al. (1981) claimed that the accuracy of EC varied considerably between herds. Because SCC in this study showed similar sensitivity and specificity between herds, SCC appeared to be a more accurate method at detecting infected quarters than EC. It was then suggested that overall, EC is unsuitable cow-side test to detect the presence of intramammary infection.

EC was also influenced by stage of lactation as well as the number of lactations (Sheldrake et al. 1983a) as the concentration of chloride increased during lactation (Caufield & Riddell 1935) and changes in EC were seen due to physiological factors in all four quarters. However, in mastitis, changes of EC occurred only in infected gland (Linzell et al. 1974). Sheldrake et al. (1983a) observed that as lactation advanced, there was little effect of EC seen in uninfected mammary quarters whereas gradual increase of EC was seen in quarters infected with minor or major pathogens.

Sheldrake et al. (1983b) found that the probability of misclassifying (Pm) quarters for EC ranged between 22 to 32%. Pm reduced the apparent distortion of percentages caused by unequal numbers of infected and uninfected quarters as it is the point at which false negative and false positive diagnoses are equal. However, this study was in conflict with Davis (1947) in which the comparison within cow by differences and ratios from different quarters yielded little advantage over absolute estimations. However, the finding was in accordance with studies by Linzell et al. (1974) and Fernando et al. (1982).

Chapter 1, 47

EC has been proposed to detect the presence of clinical mastitis earlier than by inspection by herdsmen to detect the physical changes of the milk (Milner et al. 1996). In 55% of cases, the diagnosis of mastitis by EC can be made at an average 2 milkings before any visible clinical changes occurred. In about 34% of cases, diagnosis was made at the same time as clots appeared in the milk. Thus, changes in EC as detected using automated sensors could be applied as early as or earlier than visible changes in milk detection to diagnose clinical mastitis.

Differences in the EC values between milk fractions in infected quarters have been studied by Woolford et al. (1998). In infected quarters, decreased EC values (p<0.05) were seen from first foremilk to main flow milk fractions. Meanwhile, in quarters infected with S. aureus, conductivity was increased (p<0.05) from main flow milk to stripping milk fraction. In contrast, in uninfected quarters, the conductivity decreased with successive milk fractions due to elevated fat levels. However, variation in electrical conductivity values in infected quarters was not due to fat levels but may be the result from the mixing of milk from infected to uninfected areas of the gland. Localized infection and differential drainage from the infected area of secretory cells were the factors which caused the variation change in EC from first foremilk to stripping milk fraction. The authors suggested that using the foremilk fraction provided the highest sensitivity when EC was used to diagnose subclinical intramammary infection.

Various EC traits have been used to study its association with the udder health (Norberg et al. 2003). Four EC traits defined as the inter-quarter ratio (IQR) between the highest and lowest quarter EC values, the maximum EC level for a cow, IQR between the highest and lowest quarter EC variation, and the maximum EC variation for a cow were calculated for every milking throughout lactation. All EC traits were increased markedly (P<0.05) in subclinical and clinical mastitis. The IQR performed best in classifying cows correctly with 80.6% of clinical and 45% of subclinical cases were classified correctly. About 74.8% were correctly classified as healthy. This is due to the fact that this trait reflected the level of EC rather than variation of EC.

Biggadike et al. (2002) suggested that an assessment of changes in EC should be made on an individual quarter basis to minimise other factors such as stage in lactation that influenced the changes in a whole udder. However, the authors found that the conductivity of individual quarter milk samples is not as suitable to be used as a reliable early mastitis Chapter 1, 48 predictor due to low sensitivity (54%) and positive predictor value (55%). It was suggested that cows with high EC values in their individual quarters, additional evidences are needed before initiating any treatment such as NAGase test or ATPase test.

1.2.5 Infrared Thermography (IRT)

Infrared thermography (IRT) has been employed as a diagnostic method in bovine mastitis based on heat detection generated from the udder. The thermal camera used in this method absorbs infrared radiation and based on the amount of heat generated, the images are produced and generated (Eddy et al. 2001; Mazur & Eugeniusz-Herbut 2006). The ability of IRT in detecting increased body temperature has made it possible to assess many aspects in animal industry. In animal production, it has been used to assess meat quality (Schaefer et al. 1989) and detecting oestrus in animals (Hurnik et al. 1984). Furthermore, it also can be used to assess animal welfare (Stewart et al. 2005) that is associated with stress (climate, managerial practices) (Kimmel et al. 1992; Eicher et al. 2006). Lastly, it has shown to have an advantage over other methods of diagnosis in diseases such as bovine viral diarrhoea in calves (Schaefer et al. 2004, 2007) and bovine lameness in dairy cows (Head & Dyson 2001; Nikkhah et al. 2005).

In bovine mastitis, inflammation induced by the pathogens increased the temperature of the animal. Hurnik et al. (1984) identified 4 out of 6 cases of mastitis using IRT. Endotoxin infusion into the mammary gland caused an increase of 2.3ºC of temperature as measured by the IRT (Scott et al. 2000).

Colak et al. (2008) studied the relationship between CMT score and skin surface temperature (SST) of the udder as well as the rectal temperature (RT). During mastitis, increased vascularisation at the infected site and tissue metabolism increased the SST (Berry et al. 2003) which can be detected by IRT. At cow average, strong correlation (r=0.93; P<0.0001) was seen between CMT score and udder SST but weak correlations between CMT score and RT (r=0.27; P<0.01) and between udder SST and RT (r=0.24; P<0.02). Nevertheless, IRT has shown that it was sensitive enough to detect an increase in udder temperature.

Chapter 1, 49

Although Barth (2000) claimed that IRT was not suitable to detect early stage of mastitis, Berry et al. (2003) developed a predictive model for the temperature of the udder surfaces and suggested that IRT has the potential to diagnose early stage of mastitis.

Furthermore, Hovinen et al. (2008) found an increase in udder temperature of both experimental and control quarters 4 hours post-inoculation which was in line with a rise in rectal temperature. In contrast, mild physical changes in udder and milk which occurred 2 hours post-inoculation were not detected by IRT. Swelling of the udder caused by the plasma leakage into the interstitium (McGavin & Zachary 2007) may result in undetected increases in udder surface temperature. The equal temperatures seen in both experimental and control quarters were caused by the systemic effects of LPS (Hovinen et al. 2008). Although this method might bring additional value in detecting mastitis with continuous monitoring of a herd, the authors concluded that more studies should be conducted in naturally occurring mastitis in field conditions where the systemic inflammatory reactions might be less pronounced.

Polat et al. (2010) however successfully found high predictive diagnostic ability (positive predictive value of 95.0) of IRT similar to CMT (positive predictive value of 99.2) thus making it a sensitive tool in detecting early onset of subclinical mastitis. However, several cow factors such as parity, milk production and time relative to feeding and milking cows under different environmental conditions (air temperature, humidity and velocity) should be taken into account before analysing the reliability of IRT in diagnosing early onset of subclinical mastitis.

1.3 Biomarkers in bovine mastitis The discovery of disease biomarkers has become a major research interest in recent years especially with the development of new technologies. Biomarkers are indicators of biological processes and pathological state that can reveal a variety of health and disease traits (Biomarkers Definitions Working Group, 2001). They are important analytes that can be used to predict the progress and outcome of a disease. In bovine mastitis, the search for biomarkers has been underway since the late 1990‟s. Traditional methods that are still used to diagnose bovine mastitis may have their technical drawbacks that affect their efficiency (Chagunda et al. 2006). For example, somatic cell counts (SCC) as described above appears to be a less sensitive and less specific indicator when assessing milk quality in the bulk tank milk level (Le Roux et al. 2003) than for individual cows. In contrast, Chapter 1, 50 bacteriology using culturing techniques to detect the micro-organisms that cause mastitis has disadvantages of lacking in speed, is very labour-intensive and expensive (Labohm et al. 1998; Hillerton 2000; Viguier et al. 2009). Detection of mastitis using electrical conductivity or pH is inexpensive and easy to monitor but has been shown to be relatively insensitive (Nielen et al. 1995; De Mol & Ouweltjes, 2001, Viguier et al. 2009). Furthermore, results from California Mastitis Test (CMT) can often be mis-interpreted especially when diagnosing a subclinical mastitis case that can lead to false positive or negative (Chagunda et al. 2006; Viguier et al. 2009). Thus, the discovery on specific mastitis biomarkers is needed for an early diagnosis of bovine mastitis so an early therapeutic intervention can be applied to reduce the severity of the disease.

Following mammary gland infection, various chemoattractants are released which are induced by liberation of toxins from the bacteria. These mediators include various types of cytokines attract circulating immune effector cells mainly PMN on the site of infection. The proteins of PMN have been studied for biomarker potential. The PMN cells are involved in destroying invading bacteria by releasing their intracellular granules which contain bactericidal peptides, proteins, enzymes and neutral and acidic proteases such as elastase, cathepsin G, cathepsin B and cathepsin D.

A reliable indicator has become an important interest and several criteria have been set so that a serious and prolonged period of subclinical and clinical mastitis can be avoided. A mastitic biomarker should have a fast response to infection so that the treatment course can be initiated as soon as possible. The response towards infection should also be distinguishable from the basic milk values as it is an important factor in early warning procedure. The presence of indicator only in the infected quarters and not spreading in other neighbouring quarters is essential if the intended mastitis parameter is adapted to be used at the single quarter level or composite milk level (Larsen et al. 2010).

An alteration in the permeability of the microvasculature due to inflammatory responses has led to the identification of blood proteins leaking from the intravascular compartments into milk through the separated tight junctions between endothelial and epithelium cells and have been examined as potential mastitis biomarkers. One of the better indicators is bovine serum albumin (BSA) which is one of the smaller molecular weight of protein of plasma (Giesecke & Viljoen 1974). However the determination of BSA concentrations in Chapter 1, 51 milk requires immunochemical procedures, it is difficult to automate and thus has not gained popularity.

Barta et al. (1990) investigating blastogenesis inhibitory factors suggested that mastitis might be also associated with the presence of these inhibitory factors in milk whey. Bovine lymphocyte cultures were used to detect DNA synthesis inhibitory factors in whey and showed that these factors may be released from somatic cells or other tissue cells during inflammation. In this study, clinical mastitis was associated with inhibition of Concanavalin A (ConA-) and phytohaemagglutinin P (PHA-) mitogen induced DNA synthesis by more than 95 and 86%, respectively. A positive correlation was shown between such inhibitory activity, SCC and with electrical conductivity in samples from quarters with mastitis indicating that inhibitory factors in whey may be used as a novel indicator of subclinical mastitis. However, this is relatively complicated and cannot be applied in daily practice although it can be simplified by isolation of the molecules responsible for DNA synthesis inhibition.

A study by Baeker et al. (2002) using a proteomic approach revealed the upregulation of lipocalin-type prostaglandin D synthase (L-PGDS) in quarters affected with subclinical mastitis which could be used as a marker in bovine mastitis. Prostanoids are known to have a role as inflammatory mediators via anti-inflammatory effects of this enzyme and its metabolites. Pro-inflammatory mediators which are induced by bacteria or their cell wall compartments as well as lipopolysaccharide (LPS) can activate several signalling pathways that leads to an expression of cyclooxygenase-2 (COX-2) and finally results in increased production of prostanoids. Mammary gland and mammary epithelial cells are able to produce prostanoids and leukotrienes. The upregulation of L-PGDS suggested that there is a probability of coupling of this enzyme to COX-2 induction. As a result, increases in production of prostaglandin D2 (PGD2) and its metabolites such as PGJ2, δ-12-PGJ2 and

15d-PGJ2 could be seen in the subclinically mastitic cows on this study.

A study by Zeng et al. (2009) involved determining the level of lipopolysaccharide binding protein (LBP) in naturally acquired subclinical and clinical mastitic cows in comparison with the level of acute phase protein; haptoglobin (Hp) which is becoming accepted as a diagnostic marker in bovine mastitis. However, milk and blood concentrations of LBP could not be differentiated between subclinical intramammary infection with clinically healthy cows. In contrast to LBP, blood and milk concentration of haptoglobin were Chapter 1, 52 significantly different between cows with clinical mastitis and subclinically infected as well as clinically healthy cows. Thus, LBP was not a suitable candidate in bovine mastitis and haptoglobin appeared to be a better biomarker in differentiating healthy and subclinical mastitic cows.

Another type of biomarker which has been studied to differentiate between acute and chronic inflammatory diseases is Hp-matrix metalloproteinase 9 (Hp-MMP9) complexes. Although a study by Bannikov et al. (2010) did not focus specifically on bovine mastitis, it showed that this complex which was released from neutrophils upon degranulation might serve as an important indicator during acute inflammation. This complex has shown significant difference in the serum of cattle with acute septic disease compared to those with chronic/metabolic inflammatory disease or healthy animals (Bannikov et al. 2010). However, neither individual Hp nor MMP-9 was suitable marker to differentiate between acute and chronic inflammatory diseases for example mastitis. Furthermore, it is common to find MMP-9 alone in the sera of clinically healthy animals while these proteins might have dual roles in inflammatory process either pro-inflammatory or anti-inflammatory activity. Thus, the use of Hp-MMP 9 complexes to diagnose acute septic inflammation in cattle was more reliable when compared to free Hp or MMP-9.

1.4 Acute Phase Proteins in Bovine Mastitis The measurement of acute phase proteins following infection has been used extensively in human patients as a part of monitoring disease progression and to evaluate the successful of treatment given. Following infection, a number of systemic events take place which include the acute phase response. The release of various inflammatory mediators and cytokines stimulates the hepatocytes in the liver to produce and secrete acute phase proteins (APP). The released cytokines also able to produce a wide range of host responses including pyrexia, leucocytosis, hormone alterations such as an increase in circulating cortisol and a reduction in thyroid hormone levels and trace element disturbances with reductions in the serum iron and zinc concentrations (Eckersall 1995). This is part of the innate immune response in the defence against pathogens prior to full activation of immune response.

In cattle, Hp and serum amyloid A (SAA) are the most sensitive APP during inflammatory or infectious conditions (Conner et al. 1986; Eckersall et al. 2001). Thus, evaluation of these proteins using specific assays has become one of the most promising methods to Chapter 1, 53 detect inflammation. Haptoglobin is known to be one of the most reactive acute phase proteins in bovine and exhibits high relative elevated level during acute infection. It has been shown to protect the host from oxidative activity of haemoglobin by binding to the free haemoglobin. Thus, decreased levels of haptoglobin in any sample indicate the presence of haemolysis of red blood cells. Serum amyloid A, another major APP in cattle is synthesized by the liver in different isoforms and released into the plasma following stimulation of inflammatory mediators such as interleukin-1(IL-1), interleukin-6(IL-6) and tumour necrosis factor (TNF-α).

It has been demonstrated that the concentration of Hp is low and often undetected in normal and healthy animals but can increase 50-100 times during inflammation. Although SAA is also known to be increased during the acute phase response, it seems to react faster during inflammation and produce higher differences between healthy and diseased animals compared to Hp. The use of SAA (which is a precursor of amyloid A protein) as a marker of tissue injury in cattle was originally limited due to its difficulties in purifying and quantifying as it is an apo-lipoprotein complexed in the high density lipoprotein fraction (Horadagoda et al. 1999), but commercial assays have been available for several years.

The value of using Hp and SAA as important markers in detecting inflammation has proved to be greater than the use of haematological analyses. Following on the successful measurement of the relative concentration of serum amyloid A in cattle using ELISA by Boosman et al. (1989) and purification of amyloid A from acute phase bovine serum by Horadagoda et al. (1993), the measurement of serum amyloid A (SAA) in the serum can be quantified using ELISA. Assessment of acute phase proteins is recommended compared to haematological analyses as these proteins are quite stable and the measurement can be performed on frozen samples. Haptoglobin for example, has proven to be a sensitive inflammatory indicator in both sheep and dogs (Skinner and Roberts 1994; Solter et al. 1991). These findings were supported by work in cattle by Horadagoda et al. (1999) where it was found that acute phase proteins were able to discriminate between acute and chronic inflammation. However, due to differences in the release of cytokines following infection, variations in the responses of acute phase proteins might occur. For example, the concentrations of SAA increased significantly during the early course of disease compared to haptoglobin which indicated that SAA is more sensitive to stimulation than haptoglobin. Measurements of both proteins helped to determine to which extent the inflammatory responses has occurred. Chapter 1, 54

Although these two APP are known to be elevated in bovine serum during either spontaneous or induced mastitis, studies by Eckersall et al. (2001, 2006) have shown that serum amyloid A can be produced locally in the mammary gland making it a more reliable indicator of bovine mastitis. This has suggested that measurement of acute phase proteins is more sensitive and less influenced by the physiological state of the cow. In order for the SAA to be increased in the milk during mastitis, an intramammary inflammatory stimulus is required.

One study which confirmed the presence of mammary-associated serum amyloid A3 (M- SAA3) was done by McDonald et al. (2001) in which they found high levels of this extrahepatically produced SAA3 isoform in the colostrums of healthy cows. It may have a role in providing immunological benefits to the neonates. This is further confirmed by Molenaar et al. (2009) in which they found that bovine mammary gland epithelium was capable of expressing M-SAA3 and had a role in developing mammary gland immunity.

Despite showing production of SAA in the mammary gland during mastitis, disruption to the permeability of blood-milk barrier is observed which could lead to leakage of serum proteins in milk (Eckersall et al. 2006). However, because no correlation was found between concentrations of SAA in milk and serum, they suggested that SAA might be produced locally in the inflamed mammary gland. It has been be concluded the discrepancy that exists in the levels of acute phase proteins found in serum and milk is caused by a variation in the duration and severity of the udder infection (Nielsen et al. 2004)

In one study of APP in mastitis by Kovac et al. (2007), acute phase proteins were found to be more accurately detectable in milk rather in serum and thus may be a useful tool in diagnosing mastitis and a useful marker of milk quality. Hp and SAA milk levels were significantly higher in cows with the somatic cell counts exceeded 400 000 cells/ml than in cows with SCC less than 400 000 cells/ml. Thus, it was concluded that due to high correlation of acute phase proteins levels in milk and SCC, the analysis of APP in the milk may be useful in monitoring the health of the udder.

1.4.1 Naturally occurring mastitis The dynamics of the acute phase response has been studied in naturally occurring coliform mastitis (Ohtsuka et al. 2001; Wenz et al. 2010). The changes in the level of both haptoglobin (Hp) and α1-acid glycoprotein (AGP) in serum were different in two groups of coliform mastitis; severe and moderate respectively. Serum α1-acid glycoprotein (AGP) Chapter 1, 55 was shown to be increased later in the course of disease in severe coliform mastitis cows whilst the concentration of Hp levels only peaked at 3 days after the onset of the disease (acute inflammatory condition). Thus, the severity of the inflammatory processes can affect the regulation of different APP.

A comparison study on acute phase proteins; haptoglobin, SAA and α1-acid glycoprotein

(α1-AGP) was done to evaluate the diagnostic potential of these proteins as mastitis indicators in both milk and serum (Eckersall et al. 2001). The result of this study revealed that there was no significant difference in terms of APP levels in the serum established between mild and moderate inflammation. Thus, acute phase proteins in the serum are not suitable indicator to discriminate between these groups. However, these APPs in the serum are able to differentiate between acute and chronic inflammatory condition in which Hp and SAA were both affected by acute conditions while AGP is a better indicator in chronic inflammation. AGP was not found in the milk as it is only a moderate acute phase protein in cows and that SAA and Hp might have an inherent advantage over AGP as well as α1 antitrypsin. In this study, the specificity of both SAA and Hp were 100% with no false positive results. Only SAA in milk showed significant differences between mild and moderate inflammatory responses which suggest the production of SAA in milk was under different control mechanisms from that of Hp in the milk or the SAA in serum.

Grönlund et al. (2004) also studied a naturally occurring chronic subclinical mastitis to evaluate the usefulness of acute phase proteins such as haptoglobin, serum amyloid A and adenosine triphosphate (ATP) enzyme in milk. The diagnosis of subclinical mastitis has always been challenging to diagnose due to an absence of observable clinical signs of mastitis. Although cow-side test such as CMT, measuring adenosine triphosphate or NAGase as well as electrical conductivity can be used to diagnose subclinical mastitis, these analyses are less efficient to detect subclinical mastitis and thus new disease markers and its methods of detection are needed. In this study, they found that APP level varied markedly in chronic subclinical mastitic cows but there were significant correlations between Hp, SAA and ATP. As compared to the previous study by the same author, Hp was found to be often detected than SAA. Different amount and/or combination of cytokines involved and also variation in host response towards inflammation might contribute to this result. The duration and different phases of inflammation also seem to affect the type of APP being produced. Lastly, it can be concluded that Hp and SAA analysed in udder quarter sample are preferable to evaluate the usefulness of these APP to determine the health status of udder. Chapter 1, 56

Safi et al. (2009) has proposed that M-SAA3 was the most accurate test for the diagnosis of subclinical mastitis followed by CMT, SCC, MHp, SAA and SHp. Tests on APP in milk generally were more accurate than tests on serum in the diagnosis of subclinical mastitis due to the significantly larger area under curves (AUCs) of the proteins in milk than those in serum.

Another study on naturally occurring clinical mastitis (CM) investigated the factors that influenced the concentrations of selected acute phase proteins and cytokines on mastitis caused by different pathogens (Wenz et al. 2010). They found that the level of Hp, IL-19, IL-12, TNF-α and IL-8 were greater in cows with moderate to severe versus mild systemic diseases. When analysing the season of the year when CM occurred, it appears that some proteins (APP and cytokines) increased during summer and others increased during winter. Thus, it possible that season does affect the immune response of the cows during mastitis. This result showed that with higher amount of proteins being expressed during winter, it was associated with the fact that the immunity is enhanced during short-day photo period (SDPP). Lower levels of IL-10, IL-12 and lactate dehydrogenase (LDH) in the first 60 days of milking are associated with low immunity in periparturient cows. Cows infected with gram-negative bacteria seemed to have higher level of IL-10 and IL-1β as compared to cows infected with gram-positive bacteria. Finally, this study provides a comprehensive background on the factors affecting the innate immune response of cows with naturally occurring clinical mastitis and their influences on proteins associated with infection.

1.4.2 Experimental Induced Mastitis When evaluating the acute phase response in experimental induced mastitis, it was found in an early study that serum Hp was a good prognostic indicator based on its higher level in severely affected heifers inoculated with Actinomyces pyogenes, Fusobacterium necrophorum and Peptostreptococcus indolicus (Hirvonen et al. 1996). Although acid- soluble glycoproteins (ASG) were associated with severely affected heifers, it was less sensitive than Hp as ASG is comprised of several acute phase reactants (e.g: α1-acid glycoprotein and some proteinase inhibitors). Meanwhile, the use of fibrinogen as a reliable marker was not found to be useful due to the insignificant differences between severe and moderately affected animals. As for α1-acid proteinase (α1-PI) activity, it was found not to be a suitable marker to predict the severity and outcome of the disease. Although the activity level was higher in severely affected animals, the differences are not Chapter 1, 57 significant. Fibrinogen and α1-PI however, can be used as markers to detect the presence of inflammatory condition (Hirvonen et al. 1996).

Following the revelation by Eckersall et al. (2001, 2006) on the discovery of APP in milk, a number of experimental studies have looked at the responses of the APP in milk. A study by Suojala et al. (2008) used two consecutive experimentally induced mastitis by E. coli studies to evaluate the possible carry-over effects of the previous intramammary infection. The result showed that the cows had a moderate systemic immune response towards E. coli inoculation in the first challenge but the clinical signs were milder and disappeared faster in the second challenge. The level of inflammation indicators (Hp and SAA) followed the same pattern producing significant difference in both milk and serum. Faster response towards infection and milder condition of the disease might be due to effective bacterial elimination by neutrophils through LPS recognition. Concentration of SAA in milk was higher as compared to the serum and this rapid response is due to local immune response in the mammary gland itself. However, the milk Hp concentration was only half of those in the serum and the response was less pronounced during the infection. As for the LBP, it is involved in the formation of LPS-CD14-complex which is known to increase the sensitivity of the host immune response towards gram-negative bacteria.

One experimental study on Staphylococcal causing mastitis has been performed to describe the local and systemic acute phase response during acute and chronic phases of mastitis (Grönlund et al. 2003). Staphylococcus aureus is one of major pathogen causing bovine mastitis ranging from peracute to chronic form. However, the peracute form is rare but if the acute form is not successfully treated, it can progress into chronic, subclinical form. Chronically infected cows become source of infection where the bacteria are intermittently being shed. During acute clinical mastitis, APP were increased in infected quarters suggesting the local production of proteins particularly SAA in the mammary gland. The rise of APP was correlated with the increase of SCC and the onset of clinical signs. The problem arises when diagnosing subclinical chronic form of mastitis. SAA has shown significantly higher level in infected quarters than pre-infection and healthy control quarters. Also, the increase of APP in the serum might suggest a non-specific response due to other acute inflammatory response but chronic form of mastitis was found to be able to initiate the systemic acute phase response. Thus, SAA is an excellent indicator to diagnose chronic subclinical mastitis (Grӧnlund et al. 2003).

One experimentally induced subclinical mastitis study (Eckersall et al. 2006) showed that the occurrence of increased APP in milk happened prior to the detection of increased Chapter 1, 58 serum concentrations of both of these proteins. This demonstrated that blood was unlikely to be the source of the immediate increase in milk. A moderate increase of both APP, particularly the Hp indicate that this experimental model was successful to induce subclinical mastitis in which only mild or moderate APR was stimulated. The authors also suggested that these APP can be use to estimate the severity of the mastitis condition because it showed a significant increase in APP level after second infusion of bacteria as compared to the first infusion. Results from immunocytochemistry suggested that the origin of M-SAA3 in milk may be synthesis within secretory epithelial cells and cells lining the gland cistern of the mammary gland. Although Hp synthesis was not detected by immunocytochemistry in the mammary gland following S. aureus infusion, quantitative RT-PCR has supported the probability of local production of both M-SAA3 and Hp mRNA in the mammary gland. M-SAA3 and Hp were thought to have similar activities as serum Hp and SAA and these could be useful to provide resistance to diseases as well as passive transfer of innate immunity to neonate and calf.

Another study on cows following inoculation of LPS by Larsen et al. (2010) showed that Hp level is more reliable indicator to distinguish between moderate and severe cases of mastitis. However, the acute phase reactants (Hp and SAA) have their own drawbacks for having prolonged phase of decline and that the declining phase may last for several days. The use of ELISA-based principle to analyse the milk Hp is not suitable for in-line system but the application of biosensor to determine the Hp level might give a better value on practical purposes (Åkerstedt et al. 2006). As for the concentration of enzymes involved in mastitis, only LDH, NAGase and AP are known to increase in a faster rate upon LPS infusion. They also showed clear distinct between inflamed quarters with those placebo- treated neighbouring quarters. This study also discussed the possible origin of some of the enzymes during mastitis. For example, although acid proteinase (AcP) and alkaline phosphatase (AP) are native constituents in bovine milk and the mammary gland is the source of these enzymes, enhanced expression of leukocytes lead to upregulation of these enzymes which can be seen during mastitis. Furthermore, LDH activity has shown not only originate from plasma but also from granulocytes and lymphocytes in mastitic milk. Thus, it can be concluded that AP, LDH and NAGase are the sensitive markers in mastitis as they were increased significantly as Hp, M-SAA3 and SCC during early phase of inflammation. Chapter 1, 59

1.4.3 APP and Milk Quality One study that investigated (Åkerstedt et al. 2007) the relationship between APP and SCC in all clinically healthy animals found that a large proportion of animals (53%) had detectable levels of Hp and/or SAA. Although the presence of both APP was significantly more common in milk at higher SCC, SAA was found more frequently in the milk regardless of the sample type (quarter, cow composite and bulk tank) and had a significant relationship with SCC in quarter and composite milk samples. The reason why there was no relationship between Hp and SCC in bulk tank milk might be due to the dilution effect from quarter to bulk tank level. However, although the SCC of the cow composite sample was normal, SAA seemed to be detected in all four quarters in the cow composite sample. During certain physiological conditions (dry-off and oestrus), the permeability of blood- milk barrier is altered which leads to leakage of blood proteins into milk. Thus, the presence of APP in milk suggests the independent relationship between APP and SCC.

Hp and SAA have been studied on their use to assess milk quality. Several parameters concerning milk quality such as total protein, whey protein and casein levels, casein number, proteolysis and levels of fat and lactose in cow composite sample were compared with the level of both APP (Åkerstedt et al. 2009). Healthy milk samples with detectable level of Hp showed lower total protein and casein levels but higher SCC. Healthy milk samples with detectable levels of SAA showed lower casein number, lower lactose level but also higher SCC. Although total protein level in the milk is commonly used as a milk quality parameter, casein level is a more specific quality trait since total protein may include less valuable serum proteins that may increase during mastitis. It is said that measuring casein instead of total protein level in the milk is more relevant because the degree of casein degradation also affects the processing properties of the milk other than yield and quality of the dairy products. No relationship was observed between APP and proteolysis found in this study might be due to insensitive fluorescamine method to detect small differences in proteolysis. Proteolysis might also cause decreased casein levels and lower casein numbers as proteases particularly plasmin, bind to casein and hydrolyse the proteins to soluble peptides. Lower lactose level with detectable SAA is due to the release of specific peptides during hydrolysis of casein by plasmin that may serve as regulators in lactose secretion. However, lactose is not sensitive and specific enough to be detected in the bulk tank milk and thus, not suitable to be used as milk quality parameter. Furthermore, in samples with both APP detected, reduced whey protein level was observed which might be due to reduced synthesis of the milk proteins, including the whey proteins. And lastly, Chapter 1, 60 no relationship was found between total proteins and casein levels/casein number with SCC but high SCC was found together with decreased lactose level and increased whey proteins (Åkerstedt et al. 2008).

The assessment on the composition of bulk tank milk is done to determine the quality of raw milk sample (Åkerstedt et al. 2009). Subclinical mastitis often constitutes a problem as the absence of macroscopic changes in the milk goes unnoticed and the milk from unaffected glands mixes with the milk from infected quarters. Bulk tank milk somatic cell count was reported as not a suitable predictor of raw milk quality especially for cheese production. Thus, investigation on Hp and SAA and their relationships with various raw milk quality parameters in bulk tank milk could be important to determine if the levels of APP can be related to the milk quality traits. In bulk tank milk samples with detectable Hp, lower casein content as well as lower casein number were found due to higher proteolytic activity in these samples. Proteolysis is seen in bulk tank milk but not in cow composite milk samples because of different factors. The fact that bulk tank milk consists of commingled milk from different milkings as well as milk from non-infected and subclinically infected quarters contribute to presence of proteolytic activity in bulk tank milk. Furthermore, the longer storage time before collection and freezing allows proteolysis to proceed for a longer time. Decreased levels of both APP were related to reduce lactose level as damaged epithelial cells contribute to decreased lactose synthesis. Proteolysis of β-casein released the peptides with a regulatory effect on lactose secretion (Silanikove et al. 2000). Thus, it can be concluded that APP can be serve as potential candidates to predict the raw bulk tank milk quality in relation to protein quality because the presence of APP in milk is related to disadvantageous changes in milk composition.

1.5 General Proteomics Application Proteomics has become one of the most significant „post-genomic era‟ tools due to advances in proteomic technologies which allowed an extended experimental approach to investigation of biological systems. Originally, the initial proteomic experiments which rely on two-dimensional gel electrophoresis (O‟Farrell 1975) and application of chromatography techniques (Neverova & Van Eyk 2002) could not be used for broad proteomic application due to difficulty in identifying proteins (Graham et al. 2005). However, this problem is overcome by an introduction of mass spectrometry (MS) in combination with developing genomics databases (Graham et al. 2005). Chapter 1, 61

Proteomics is defined as the study of the proteome; a full complement of expressed proteins from the genome of a given tissue, cell or biological fluid in all isoforms, polymorphisms and post-translational modifications (PTMs) at a particular point of time. (Wilkins et al. 1995; Graham et al. 2005; Manso et al. 2005; Hiendleder et al. 2005; Mitton & Kranias 2003). Proteomics provides a holistic and global approach to separate, identify, characterize and quantify proteins and their PTMs to provide information on protein abundance, location, modification and protein-protein interaction in a proteome of a given biological system (He & Chiu 2003; Stoughton & Friend 2005). Principally, proteomics offers an inventory of the different proteins and their modifications within a defined biological state and comparison of proteomes derived from different biological states, e.g. normal in comparison to diseased states or with treatment to non-treatment (Hiendleder et al. 2005).

According to Mitton & Kranias (2003), proteomics has advantages over genomics and transcriptomics because cellular protein composition and activity are positively correlated with changes in overall cellular function. Furthermore, proteomic methods have the potential to generate a bulk of data quickly and they are able to generate high-throughput systems which are useful to analyze a complex biological system. Furthermore, proteomics is able to determine the time course of protein behaviour and to characterize the presence and relevance of PTMs to a given process.

While its potential usage is outstanding, proteomics faces three major challenges (Mitton & Kranias 2003). First, cellular proteomes are highly dynamic with changes in the proteome occur at the same time as changes in cellular functions. Thus, great care needs to be taken to standardize and interpret the results. Second, proteomics has to deal with a variety of proteins in terms of its molecular weights, partition coefficients (hydrophobicity/hydrophilicity), electrostatic charges, ion chelation properties and PTMs (glycosylation, lipid conjugation and phosphorylation). This also complicates the detection of low-abundance proteins (Hiendleder et al. 2005). Finally, technically difficult and time- consuming proteomics methods have been considered as a major hurdle although vast improvements in the last decade have improved its reliability and efficiency.

1.5.1 Proteomics in Animal Sciences The use of proteomic methods in animal sciences can help researchers to identify key proteins and their changed expression in response to different experimental conditions. Furthermore, it will provide data libraries of the abundant proteins expressed in a single Chapter 1, 62 cell, cell compartments, tissues and body fluids (Lippolis & Reinhardt 2005; Cho et al. 2006; Reinhardt & Lippolis 2006; Radosevich et al. 2007). These proteins derived from pathogens or hosts are important to examine the status of disease progression and recovery, organ development or host responses to stress. Furthermore, it also helps with the intervention of diagnostic tests, development of new therapies or even on a discovery of disease biomarkers (Lippolis & Reinhardt 2008).

With the technologies of proteomics promised to provide the identification of proteins that are differentially expressed in normal and disease conditions, naturally occurring animal diseases can be studied to improve the health of animals which in turn may lead to improvements in the characterization, clinical management and treatment of human diseases through comparative medicines (Doherty et al. 2008).

For example, dilated cardiomyopathy which occurs in the Simmental-Red Holstein crossbreed of cattle has been studied to investigate its molecular pathogenesis (Weekes et al. 1999). This study revealed that poly-ubiquitination of proteins was increased in this disease and may play a role in heart disease. Paratuberculosis or Johne‟s disease has been implicated as a possible factor in the development of Crohn‟s disease in humans (Grant 2005). Ten proteins have been identified as up-regulated proteins in naturally occurring Johne‟s disease which provide an important fact for the host-parasite relationship which can act as model system for the parallel Crohn‟s disease in human (Hughes et al. 2007).

Proteomics has been applied to study the innate immunology pathways in stress-induced immunosuppression (Lippolis et al. 2006a; Mitchell et al. 2007). One of the bovine innate immunity proteins discovered was histone H2A.1 in bovine milk (Lippolis et al. 2006b) which involved in neutrophil killing mechanism (NET) (Brinkmann et al. 2004). Additionally, proteomic investigation on milk fat globule membrane proteome (MFGM) revealed that mammary epithelial cells may be directly involved in signalling the early recruitment of innate immune cells due to the presence of innate immune signalling molecules, TLR2, TLR4 and CD 14 (Reinhardt & Lippolis 2006). This study on MFGM also provides new data on mammary development in early lactation.

Proteomic analysis on bacterial-causing diseases helps in developing new diagnostic test from new protein markers and also development of vaccine from newly identified candidate proteins. Examples of animal diseases which used similar proteomic approaches are Johne‟s disease (Radosevich et al. 2007), Brucella abortus (Connolly et al. 2006), Chapter 1, 63

Pasteurella multocida (Boyce et al. 2006) and Mycobacterium bovis (Mollenkopf et al. 2004).

The proteomes from 3 important bacteria used in the production of dairy products; Lactobacillus bulgaricus, Lactococcus lactis and Streptococcus thermophiles have been studied for any protein changes resulting from growth in laboratory medium or in milk (Rechinger et al. 2000; Derzelle et al. 2005; Gitton et al. 2005). This data provides new information to food scientists in modifying these bacteria to achieve good food production practices.

Proteomic approaches also been widely used to examine post mortem changes in slaughtered beef and pork that affect the quality of the meat (Jia et al. 2006a,b; Sayd et al. 2006; Laville et al. 2007; Suman et al. 2007). In addition, the variation in types and amounts or proteins in seminal fluid of bull has been studied using proteomic methods (Kelly et al. 2006). Although it is not known yet if these findings will lead to improvement of bull fertility, it is likely that such methods will have a good future in animal production profitability.

Because proteomics is able to screen large numbers of proteins simultaneously, it has become an important role at identifying disease biomarkers from the thousands of molecules present in a tissue of body fluid (Ilyin et al. 2004; Seo & Ginsberg 2005). In order for proteomics to be used effectively in biomarker discovery, a combination of efficient and stringent separation technologies and high-resolution mass spectrometry (Aebersold & Mann 2003; Rifai et al. 2006) are needed. Cairoli et al. (2006) applied proteomics to study the changes in protein expression in pregnant cows and in the peripartum periods in cows with or without postpartum uterine infections. They found that both haptoglobin and orosomucoid/α1-acid glycoprotein fluctuated at the time of calving. Orosomucoid is also known to significantly decrease in cows with endometritis compared with healthy cows which hold promise as a prognostic biomarker in cattle health postpartum. Proteomics has also identified PR-39 in pigs with respiratory disease as a potential disease biomarker at day 21 of infection (Hennig-Pauka et al. 2006).

1.5.2 Milk Proteomics Although milk proteins have been studied for well over 50 years, only recently have proteomic techniques been widely used to study milk and milk products by separation of major milk proteins including caseins (αs1-, αs2-, β- and κ-casein) and whey proteins (α- Chapter 1, 64 lactalbumin, β-lactoglobulin and bovine serum albumin) (O‟Donnell et al. 2004; Manso et al. 2005).

The milk proteome is an extremely complex entity due to the presence of several molecular forms of many of the proteins as a consequence of different genetic variants, changes in the level of phosphorylation or glycosylation as well as the localization of PTMs of milk proteins (Eigel et al. 1984). Furthermore, there are a number of lower abundance proteins such as lactoferrin, immunoglobulins, glycoproteins, hormones and endogenous enzymes (Fox & Kelly 2003) present in the milk which may also be studied by means of proteomic approaches.

An early observation on bovine and human milk proteins was first reported by Anderson et al. (1982) using high resolution 2 DE. Major milk proteins were demonstrated in this 2D mapping study which included α-LA, β2-LG, α- and β-caseins, 4 post-transitionally modified κ-chains, para κ-casein, Lf and albumin. Studies by Galvani et al. (2000) performed on powdered and fluid milk samples using MALDI analyses revealed the electrophoretic position of common high abundance proteins (caseins, β-lactoglobulin, etc) that can be found in milk and with modifications such as proteolysis, glycosylation and phosphorylation.

Normally, low-abundance proteins in milk are rarely seen on traditional 2DE maps due to high amount of major proteins in milk which obscure the detection of low-abundance proteins. Incomplete solubilization during sample preparation can also reduce the detection of low abundance proteins (Manso et al. 2005). However, low-abundance milk proteins account for approximately 5% of the total milk proteins and a number of studies have been applied on these proteins to aid in the search of their biological role and impacts on animal and human health (Lippolis & Reinhardt 2005; Reinhardt & Lippolis 2006; Fong et al. 2008; Smolenski et al. 2007; Reinhardt & Lippolis 2008). Low abundance proteins are frequently thought to have an important role as regulatory proteins, signal transduction proteins, receptors and enzymes (Fox & Kelly 2003).

An early study of low-abundance milk proteins was done using human colostrum and mature milk samples (Murakami et al. 1998). In total, 22 low-abundance proteins were identified using 2DE followed by microsequencing in both human colostrum and mature milk samples. Interestingly, no significant difference was found on 2DE patterns of low- abundance human milk proteins between colostrum and mature milk samples. Chapter 1, 65

Further study on milk proteins was carried out by Yamada et al. (2002) using the same proteomic methods (2DE and microsequencing) but with bovine colostrum and mature milk. In this study, 29 low-abundance proteins were identified with several proteins being present only in bovine colostrum such as gelsolin, chitinase3-like 1, apolipoprotein H, complement C3 and fibrinogen β-chain. There may be important in the health and development of calves during early lactation. They also observed significant difference in 2DE patterns of low-abundance proteins between bovine colostrum and mature milk samples.

Another study by Baeker et al. (2002) demonstrated the expression of a group of proteins from the whey from inflamed mammary quarters. These proteins were identified by MALDI-TOF as isoforms of lipocalin-type prostaglandin D synthase and suggested to be a new biomarker of bovine mastitis.

Low-abundance proteins in human colostrum were studied by Palmer et al. (2006) using 2D-LC MS/MS, who found around 151 proteins of which 83 proteins were not previously reported. In this study, immunoabsorbents were applied to remove the major proteins: secretory IgA, lactoferrin, alpha-lactalbumin and human serum albumin (HSA). Most of the newly identified proteins exhibited many important functions for the newborn‟s immunity, growth and nutrition such as complement C1r-like proteinase, complement factor I, calgranulin A, calgranulin B, cysteine-rich secretory protein-3, mucin 5B, neutrophil gelatinase-associated lipocalin, and uteroglobinrelated protein 2.

The whey proteins are the soluble part of protein fractions. Major components of whey proteins include Igs, α-LA, β-LG, lactoferrin (Lf), serum albumin and lactoperoxidase (Lp) (Severin & Wenshui 2005). According to Casado et al. (2009), whey proteins can be further separated using ultrafiltration to concentrate the proteins or generates whey protein concentrates (WPC) respectively.

Proteomic approaches have been used to study whey proteins particularly in relation to the beneficial effects of these whey proteins and their bioactive peptides as functional ingredients in food (Michaelidou & Steijns 2006; Toba et al. 2000). Furthermore, many low-abundance proteins which arise from proteolytic activity from the major milk proteins, extrusion or leakage from the plasma or mammary epithelia (Hogarth et al. 2004; Larsen et al. 2006) or are products of the somatic cells (Kelly & McSweeney, 2003 ; Paape et al. 2003) might contribute to the health benefits of whey proteins. Chapter 1, 66

Various fractionation methods have been employed to increase the resolution of such low- abundance proteins in whey fraction such as isoelectric focusing (IEF) (Zuo & Speicher 2002), affinity tagging (Holland et al. 2006) and immunoabsorption (Yamada et al. 2002). In one study by Fong et al. (2008), the use of semi-coupled anion and cation exchange chromatography to fractionate whey from late-lactation milk samples into acidic, basic and non-bound fractions has revealed a large number of minor whey proteins which have not been found previously in bovine milk such as factor XIIa inhibitor, tetranectin, CD5 antigen-like, complement components C4, C8, C9, factors I and D as well as nucleobindins.

In a recent work by D‟Amato et al. (2009), the use of a combinatorial peptide ligand library (CPLL) has been shown to enhance the low-abundance protein species in milk. This work which comprised of IEF in gel followed by immunoblotting with unambiguous detection of polymorphic IgG revealed 100 unique gene products in milk which were not previously mentioned. In total, about 149 proteins were presented in a comprehensive whey proteins list under a combination of untreated and treated colostrum and mature milk with CPLL1 and CPLL2 (library 1 and 2 respectively).

However, Le et al. (2011) demonstrated that by using ion exchange to fractionate milk proteins, higher sample load was allowed which in turn, allowed more whey proteins to be identified. The total of whey proteins found was double the number of proteins identified using CPLL with 293 proteins found in bovine colostrum and mature milk. About 217 proteins were found in both bovine colostrum and mature milk whilst 36 proteins were only found in colostrum and 40 proteins came only from mature milk. Semi-quantitative analysis also revealed several lower-abundance proteins being up-regulated in colostrum such as CLU (complement system), MFGE8 (antiviral activity), CATHL1 (antibacterial activity), SAA3 (acute-phase protein), CD5L (immune system), PGLYRP1 (innate immunity and antimicrobial) and CFB (complement system) that function as natural defence mechanism.

A study on human whey proteome was performed by Liao et al. (2011) using LC-MS/MS with enrichment methods using Proteominer Protein Enrichment Technology. Major whey proteins such as Lf and α-LA or even casein subunits were still detected when using this approach but a total of 115 proteins have been identified of which 38 proteins not being mentioned before in human milk. These proteins were known to perform a wide variety of physiological functions in milk. Relative quantification of these minor proteins also showed that α -1-antitrypsin, α –lactalbumin, carbonic anhydrase 6, chordin-like protein 2, Chapter 1, 67 galectin-3-binding protein, lactadherin, lactoferrin, prolactin-inducible protein, and tenascin were significantly expressed in higher levels in both colostrum and 1-month milk whey with immune modulating functions.

Proteomic studies on milk fat globule membrane (MFGM) provide an understanding on lipid synthesis and secretion in the milk. By characterizing its proteins associated with this membrane, one may be provided with information on signalling and secretory pathways of the mammary gland and eventually the cellular origin of the membrane (Casado et al. 2009). However, MFGM is known to be highly hydrophobic and thus, solubilization is difficult. To overcome such problems, various solubilization strategies (Quaranta et al. 2001; Cavaletto et al. 2002) have been employed including the use of detergent or reducing agent solution as well as by extraction using organic solvent that may help to improve the recovery of MFGM proteome (O‟Donnell et al. 2004).

In early studies, MFGM proteins were identified as mucins, xanthine oxidase and butyrophilin) by immunological methods and N-terminal sequencing (Cavaletto et al. 2004). Mather et al. (2000) also identified MFGM proteins using traditional biochemical methods which were slow, laborious and addressed only one protein at one particular time. In humans, multiple forms of both butyrophilin and lactadherin were found after combining SDS-PAGE with sequence analysis (Cavaletto et al. 1999). Meanwhile, in bovine, major proteins associated with MFGM were demonstrated by Mather (2000) which includes mucin 1, xanthine oxidase, CD36, butyrophilin, adipophilin, lactadeherin and fatty acid binding protein.

Several studies on human MFGM proteins have been investigated. According to Quaranta et al. (2001), about 23 proteins identified by microsequencing and MS analysis confirming the presence of lactadherin and butyrophilin as the major glycoproteins in human MFGM. Other proteins that have been revealed include adipophilin, disulfide isomerase ER-60, apolipoprotein J as well as carbonic anhydrase which were not previously mentioned in human milk.

Charlwood et al. (2002) further demonstrated the presence of previously resolved MFGM proteins which were α-LA, Lf, lysozyme precursor, β-CN, clusterin, milk fat globule EGF- factor 8 protein and polymeric IG receptor precursor. By characterizing the N-linked glycans found in MFGM proteins, multiple fucosylation products can be studied which give rise to maternal preparation to the infant‟s immune system. Chapter 1, 68

The development of 2-DE database for human MFGM proteins has led to the characterization and identification of 107 protein spots with multiple PTMs on lactadherin, butyrophilin, Lf, adipophilin and carbonic anhydrase. Of these proteins, 46% were known previously as MFGM proteins composition, 10% involved in folding and protein destination and another 9% involved in intracellular transport and/or receptor activity. Also, the discovery has been made of low-abundance proteins in MFGM which involved in cell development and differentiation, cell motility and signal transduction.

Reinhardt & Lippolis (2006) first reported the proteomic study on bovine MFGM which aimed to provide a good research foundation on the secretory mechanism of mammary gland in this species. They applied an extraction of MFGM using carbonate to enrich the intrinsic membrane proteins of MFGM and found 120 proteins from bovine MFGM. Most of these proteins were associated with membrane/protein trafficking or cell signalling. The most significant findings of this study were the presence of CD14, TLR2, and TLR4 on MFGM suggesting a role of protecting the mammary gland against infection.

Later, Reinhardt & Lippolis (2008) used a more advanced proteomic approach to identify and measure the level of protein expression changes in MFGM from colostrum to milk on day 7 of lactation. An amine-reactive isobaric tag (iTRAQ) was used to quantify protein changes in MFGM between colostrum and day 7 milk and revealed 138 MFGM proteins in which 26 proteins were upregulated whilst 19 proteins downregulated in d 7 MFGM compared with colostrum MFGM. Amongst proteins that were significantly upregulated individually in d 7 MFGM were mucin 1 and 15, the tripartite complex of proteins of adipophilin, butyrophilin, and xanthine dehydrogenase as well as proteins involved in lipid transport synthesis and secretion (acyl-CoA synthetase, lanosterol synthase, lysophosphatidic acid acyltransferase, and fatty acid binding protein).

Vanderghem et al. (2008) have established a simple and fast method to purify the MFGM by using Zwitterionic detergents (4% (w/v) CHAPS). This method proved to remove the majority of skim milk proteins as no caseins were found on their 2-DE gel. A total of 230 proteins spots were excised and analysed by MALDI-TOF/TOF and further characterized by PMF and MS/MS. Amongst the proteins identified were butyrophilin, lactadherin, adipophilin, xanthine dehydrogenase/oxidase and fatty acid binding proteins as previously mentioned.

Over the past few years, an increasing interest in quantification of proteins in MFGM has been seen. This is due to the important bioactivity of MFGM proteins in different health Chapter 1, 69 benefits in human such as anti-cancer effects (Spitsberg & Gorewit 2002), anti-bacterial effects (Schroten et al. 1992; Wang et al. 2001) and anti-viral effects (Murgiano et al. 2009). Thus, Fong & Norris (2009) have used the absolute quantification (AQUA) technique to quantify the 6 major MFGM proteins in butter milk protein concentrate. The released peptides were measured at the same time using liquid chromatography-high- resolution selected reaction monitoring-mass spectrometry (LC-HSRM-MS).

Due to the importance of the MFGM efforts have been made to increase the efficiency of sample preparation in the protein extraction and solubilization prior to proteomic analysis of MFGM protein, Bianchi et al. (2009) compared 4 different types of lysis buffers with different detergents and chaotropes together with delipidation to maximize the visualization of MFG protein expression profile. In conclusion, delipidation, urea/thiourea combination and CHAPS/TritonX100 application as surfactants resulted in a proper MFG protein pattern visualization. In all, 82 protein spots with 42 different polypeptides were identified by MS and a reference 2-D map of cow MFGM was produced in this study. The majority of proteins identified were involved in lipid synthesis or in fat globule secretion.

Affolter et al. (2010) have used 3 different qualitative and quantitative analytical profiling on two MFGM enriched milk fractions (whey protein concentrate; WPC and buttermilk concentrate; BMC). An LC-MS/MS-based shotgun approach identified 244 proteins in WPC and 133 in BMP with 4 main functional categories of proteins (signal transduction, proteolysis, immunity and defence and lipid metabolism). A second approach using label- free profiling gave a rapid and efficient semi-quantitative comparison of the top 34 proteins present in both fractions. Lastly, the application of stable isotope dilution combined with LC-MRM-MS/MS technology to quantify proteins gave a highly flexible and adaptable analysis for a large number of proteins in a complex mixture.

The complexity of milk in its protein heterogeneity arises from significant levels of phosphorylation and glycosylation together with genetic variation, proteomic analyses have been used for the discovery, detection and characterization of these post translational modifications (Baeker et al. 2002; Charlwood et al. 2002). The majority of caseins exist as mixture of phosphoproteins. Mamone et al. (2003) identified more than 30 phosphorylated caseins using combined polyacrylamide gel electrophoresis with MALDI-TOF and nanoelectrospray-tandem mass spectrometry (nESI-MS/MS). Holland et al. (2004) in other hand identified 10 forms of κ-casein at different isoelectric point (pI) values as well as two phosphorylated and glycosylated forms. Chapter 1, 70

Glycosylation also occurs in numerous milk proteins particularly on both N- and O-linked glycoproteins. As described by Holland et al. (2004), The O-glycosylation sites were identified by tandem MS with 10 different glycoforms of κ-CN were resolved and identified. Charlwood et al. (2002) on the other hand studied the N-linked glycosylation on four major proteins of the human MFGM namely clusterin, lactotransferrin, polymeric immunoglobulin receptor and lactadherin which possess a wide range of different sugar motifs. Proteins were separated and identified by MALDI-TOF analysis while Tandem MS and nESI source were used to identify the sugar constituents of isolated glycans.

1.5.3 Proteomics on Bovine Mastitis Changes in the milk proteome during bovine mastitis could provide a significant insight on variation in protein composition that could lead to the discovery of protein biomarkers for bovine mastitis. Furthermore, by applying proteomic analyses, it may be possible to identify specific and potential protein target for therapies and immune manipulations against bovine mastitis (Yong et al. 2009).

Hogarth et al. (2004) applied pre-fractionation to remove abundant caseins from milk so that proteomic analyses of mastitic whey in comparison with normal whey could be investigated. However, despite an attempt to remove caseins, many other higher abundance proteins were seen from the results of this study and few low abundance proteins were apparent suggesting an alternative approach to remove these high abundance proteins. Serum albumin and serotransferrin were among the proteins found to be present or increased in mastitic whey. It is interesting to see that normal major whey proteins such as β-LG, α-LA and the casein remnants were absent from whey samples from cases of clinical mastitis.

Smolenski et al. (2007) analyzed mastitic milk for the presence of minor milk proteins involved in host defence mechanism. By using direct liquid chromatography-tandem mass spectrometry (LC-MS/MS), and two-dimensional electrophoresis (2-DE) followed by matrix-assisted laser desorption ionization-time-of-flight (MALDI-TOF) mass spectrometry (MS), several proteins involved in host defence were identified in bovine milk such as chaparonins, proteins secreted from neutrophils, lymphocyte proteins, DNA- binding proteins, enzymes, transport proteins and structural proteins probably present as a consequence of the shedding and subsequent lysis of somatic cells in milk. A number of host defence proteins were also found in MFGM and there were significant differences in the MFGM proteins between mastitic and healthy milk. Chapter 1, 71

Furthermore, Boehmer et al. (2008) successfully identified complement factors C3 and C4 and the apolipoproteins A-IT and C-Ill in mastitic whey samples inoculated with E. coli. It has proved that by using 2-DE and MALDI-TOF PSD, the identification of low-abundance proteins can be achieved without prior sample clean-up. In addition, a wide range of differentially expressed proteins were identified in mastitis whey compared with normal whey due to larger sample size which improve the discovery of protein biomarkers in coliform mastitis.

Proteomic approaches have also been applied to study protein expression of the mammary gland from both normal and clinical mastitic dairy cows as done by Yang et al. (2009). Six differentially expressed proteins were identified by tandem MS such as haemoglobin-beta, κ-CN and Tryptophanyl-tRNA-synthetase (TrpRS) which can be found from healthy mammary gland tissues as well as cytochrome C oxidase and annexin V which were identified from mastitic mammary gland tissues.

Over the years, an interest on quantification of the abundance of proteins exists in a complex biological fluid has grown tremendously. In milk, Boehmer et al. (2010) have applied an LC-MS/MS label-free approach to quantify relative changes in identified proteins over a time course following E. coli infection of the udder. A number of medium- to high-abundance proteins such as αS1-, αS2- β-, and κ-caseins and whey proteins serum albumin, β-lactoglobulin, and α-lactalbumin were found in all whey samples at all time points. In addition to that, lower abundance proteins of potential use as inflammation markers were also characterized such as lactoferrin, transferrin, apolipoprotein AI, fibrinogen, glycosylation-dependent cell adhesion molecule-1, peptidoglycan recognition receptor protein and cyclic dodecapeptide-1.

Danielsen et al. (2010) have also applied quantitative proteomic approaches to assess the protein differential expression in the milk proteome of individual cows before and immediately after LPS challenge. In all, 49 proteins were found to be significantly upregulated during LPS challenge including apolipoproteins, complement C3 and C4 as well as other anti-inflammatory proteins.

Pathway analysis can be incorporated when studying the enriched proteome that emerged during E. coli and S. aureus responses as done by Ibeagha-Awemu et al. (2010). The proteomic profiles of whey milk from naturally infected E. coli and S. aureus in comparison to whey from healthy cows were revealed by one-dimensional, liquid chromatography tandem MS (LC-MS/MS). A total of 73 proteins were found to be Chapter 1, 72 differentially expressed between normal and whey from quarters infected with E. coli appeared to cause more significant alteration in the concentration of the affected proteins than S. aureus itself. The acute phase response signalling pathway was the most significant enriched canonical pathway in this study. In addition, more lower abundance proteins were identified as compared to other previous studies such as the serpins and proteins with binding activities (glucose phosphate isomerizes and PFN1)

Proteolysis of casein which is known to increase in bovine milk during mastitis has been studied by Larsen et al. (2010) after infusion of lipoteichoic acid from S. aureus.

According to 2-D gel electrophoretograms, hydrolysis of β- and αS1-caseins occurred in milk 6 hours after infusion with the toxin. Those peptides were subsequently separated by reversed-phase HPLC and C18 columns and further identified by MALDI-TOF MS/MS. About 20 different peptides were successfully characterized and confirmed to originate from αS1- and β-CN.

Proteomic study on different pathogens causing bovine mastitis could also provide important data on the proteome of bacteria isolated from mastitis cases. This could aid in defining specific vaccine and therapeutic targets as suggested by Taverna et al. (2007). They produced a (2-DE) reference map of surface associated proteins of isolated S. aureus from a case of bovine mastitis with protein identification using MALDI-TOF MS.

1.6 General Peptidomics Peptidomics is regarded as the sequel to the proteomic technologies and is one of the new branches of chemical biology based on post-genomic technological platforms. The term peptidome refers to the whole set of endogenous peptides present in a given biological sample (Ivanov et al. 2009) and peptidomics is a comprehensive study of all natural peptides in an organism, a tissue or a body fluid (Clynen et al. 2001; Schrader & Schultz- Knappe 2001). In the previous discussion on proteomics, 2D-PAGE is often utilized to separate and visualize the protein complement present in any cells, tissues and fluids followed by identification of expressed proteins by the MS (Baggerman et al. 2004). However, due to major limitation of 2D-PAGE in detecting molecules between 10kDa and 200kDa, these standard proteomics approaches are unsuitable for the analysis of peptidomes (Ramström & Bergquist 2004). It is thought that the very important groups of peptides and neuropeptides are outside the range of this mass region (Baggerman et al. 2004). Furthermore, according to Schulte et al. (2005), due to high complexity and concentration of other proteins in a sample, 2D-PAGE is not an appropriate method to Chapter 1, 73 detect these important endogenous peptides especially as they are often present at very low concentrations.

Peptides are referred to as low-molecular weight polypeptides molecules with the mass of <20kDa (Ramström & Bergquist 2004, Hu et al. 2009). They are often the products of proteins after proteolytic cleavage of the protein precursor by active proteases that have very specific functions as mediators and regulators of many physiological processes in nervous, endocrine, immune and other systems (Ivanov et al. 2009, Hu et al. 2009). Peptides just like proteins change dynamically according to different physiological and pathological conditions and are influenced by metabolism, disease states, stress, drug intervention, genetic modifications and other external factors (Ivanov et al. 2009; Ramström & Bergquist 2004). Although the circulating peptides in the body fluid or tissue may reflect the biological events, the real function of specific peptides is frequently unknown and further studies were needed (Petricoin et al. 2006). According to Hu et al. (2009), peptides in the peptidomes are divided into two groups; first, the bioactive peptides that have important roles in biological processes such as cytokines, hormones, growth factors and MHC class 1 peptides; and second, the degraded peptides arising from proteolytic cleavage.

Peptidomics was only introduced at early 2000s after several advances in the development of MS techniques coupled with the completion of the genome sequences in majority of living species (Ivanov et al. 2009). Subsequently, several fields of studies have successfully employed peptidomic approach such as neuroendocrine studies (Baggerman et al. 2002; Clynen et al. 2003), biomarker discovery (Yuan & Desiderio 2005; Schrader & Salle 2006) and therapeutic development against oncology, neurodegeneration, osteoporosis and others (Clynen et al. 2008).

In earlier years, the study of peptides involved the purification of peptides from tissue extracts with the extracted peptides being characterized by Edman sequencing (Ramström & Bergquist 2004). As a result of the introduction of MS, hundreds of newly identified peptides were discovered and sequenced (Ivanov et al. 2009; Ivanov et al. 2005) and termed as peptide profiles (Slemmon et al. 1997) or peptide pools (Karelin et al. 1998). It is not an easy task to analyse peptides in a real sample due to their low concentrations, the complexity of the sample together with the presence of other components in the sample in a wide range of concentration and the availability of the analytes at different chemical properties (hydrophobicity, size, pI) (Ramström & Bergquist 2004). Thus, a very sensitive, unbiased method should be applied with short running time, and which is easily automated Chapter 1, 74 and able to handle a great abundance of data. The different peptidomics approaches will be discussed in the final chapter.

1.6.1 The presence of peptides in bovine milk and dairy products Milk or dairy products possess an important role in providing the main nitrogenous nutrients in human food as it has a high content of proteins. Many breakdown products of proteins upon ingestion of milk proteins are called bioactive peptides (Srinivas & Prakash 2010; Schmelzer et al. 2004). The hydrolytic reactions of digestive enzymes present in the gastrointestinal system as well as in vitro proteolysis by proteolytic enzymes such as plasmin, (Bastian et al. 1991), cysteine proteases or acid proteases (Wedholm et al. 2008) lead to the production of these bioactive peptides. Bioactive peptides in milk or dairy products such as cheese and yoghurt provide a health enhancing factor that can be incorporated into functional foods and /or used as nutraceuticals (Meisel & Fitzgerald 2003). Most of the bioactivities of these peptides remain in a latent state within the protein sequence and to exert their functions, proteolytic processes must be initiated (Smacchi & Gobbetti 2000; Srivinas & Prakash 2010). However, some of bioactive substances in milk are fully active as proteins (Smacchi & Gobbetti 2000). These bioactive peptides in dairy products serve numerous biological and physiological functions.

The functions of bioactive peptides in dairy products have been reviewed by Smacchi & Gobbetti (2000) which include opioid characteristics, antihypertensive, antimicrobial and antithrombotic effects. Examples of these peptides include peptides from αS1- and β-CN (Maruyama & Suzuki 1982), β-lactoglobulin (Mullaly et al. 1996), α-lactalbumin (Parker et al. 1984), and lactoferricin (Yamauchi et al. 1993). Some of these milk-derived peptides exert multi-functional activities such as β-casomorphins and casokinins; both are Angiotensin-1 Converting Enzyme (ACE)-inhibitory and immunostimulatory (Grant et al. 1993; Meisel & Schlimme 1994).

Smacchi & Gobbetti (2000) have also produced lists of bioactive peptides present in different types of dairy products such as milk and milk hydrolysates, fermented milks and cheese. Proteolytic bioactivities depended upon the differentiation, functional states as well as the intramammary inflammation of mammary gland (Schanbacher et al. 1997). Various peptides were also found in fermented milk which influenced by the type of bacteria starter used. The production of bioactive and non-bioactive peptides during proteolytic event of cheese ripening may inhibit the lactic acid bacteria (LAB) peptidases in many ways. Chapter 1, 75

In another study, Srivinas & Prakash (2010) have isolated and characterized the multifunctional properties of bioactive peptides from α-CN from bovine milk. To generate peptides from α-CN, several proteolytic enzymes were employed such as aminopeptidase, carboxypeptidase, fungal and bacterial proteases and chymotrypsin. However, chymotrypsin hydrolysates were chosen as they released smaller size of peptides. Peptides identified in this study showed multifunctional properties such as ACE inhibition, antioxidant activity, propyl endopeptidase inhibitory activities and antimicrobial properties.

In terms of technical determination and characterization of proteolytic products of dairy products, numerous methods have been employed. In early years, these include polyacrylamide gels with urea (urea-PAGE), denaturing agent (SDS-PAGE) or IEF techniques (Molina et al. 2000). With time, the separation and identification of peptides has been successfully performed by coupling these previous analytical techniques with the MS. Details on the various analytical tools on peptide studies will be discussed in the final chapter.

Most of the studies on bovine milk peptides revolved around proteolytic activities on casein, the major milk protein. Plasmin has been regarded as the main indigenous proteinase in bovine milk (Bastian et al. 1991; Somma et al. 2008) and it preferentially cleaves Lys-X and Arg-X residues (Le Bars & Gripon 1989). This enzyme activity increases with the stage of lactation, lactation number and severity of bovine mastitis (Politis et al. 1989; Schar & Funke 1986) and it leads to detrimental effects on milk quality due to decreased amount of intact CN.

Recio et al. (1997) have employed capillary electrophoresis (CE) with hydrophilically coated capillary and low pH buffer containing urea to separate proteolytic activites of plasmin and chymosin on caseins. They identified several casein breakdown products which include γ1-CN A1, γ1-CN A2, γ1-CN B, γ1-CN C, γ2- CN A, γ1- CN A3, γ2- CN B,

γ3- CN A and γ3- CN B including proteose peptones derived from different genetic variants of β-CN. Breakdown products of αS1-CN and κ-CN were also found which were produced by chymosin proteolysis.

Molina et al. (2000) employed CE to study the fractions of casein and its breakdown products treated with plasmin and chymosin from three different species of animals (cow, ewe and goat). It showed that hydrolysis of β-casein by plasmin gives rise to γ1-CN, γ2-CN and γ3-CN. Additionally, they found a decrease in the peaks corresponding to αS2-CN as it Chapter 1, 76 was also very susceptible to hydrolyse by plasmin due to its high lysine residues (Le Bars & Gripon 1989). Hydrolysis of casein in cheese by chymosin and rennet was also studied with its main degradation products include para-κ-CN and αS1-I-CN.

Another study on peptides derived from bovine β-CN by peptic cleavage was studied by Schmelzer et al. (2004). By using two types of MS analysis; MALDI-TOF MS with post- decay source and LC-MS/MS, they successfully identified 41 peptides with a high sequence coverage (75%). The peptides identified had lengths of between 2 and 36 amino acids which originated from the C-terminus of the protein.

Somma et al. (2008) compared two different approaches between capillary isoelectric focusing (cIEF) and ultra-thin layer isoelectric focusing (ultIEF) in combination with MS to characterize the products of β-CN digestion by plasmin from both bovine and water buffalo milk samples. The presence of a novel CN breakdown product was found from the hydrolysis of water buffalo β-CN which was not present in hydrolysis of bovine β-CN. They concluded that cIEF is faster, requires shorter processing times and does not involve staining when compared with ultIEF technique.

Apart from proteolytic activities, studies of peptides have also been employed to investigate the presence of adulteration in milk or other dairy products. The most common fraud observed was the addition of cow‟s or sheep‟s milk to goat‟s cheese due to seasonal fluctuations, availability and price (Guarino et al. 2010). Thus, the need of rapid, fast and reliable control methods to accurately determine the level of adulteration in milk or dairy products is increasing. Previously, the use of immunological technique (Levieux & Venien 1994), PCR-techniques (Bottero et al. 2003) or slab gel electrophoresis (Addeo et al. 1990) have been used to detect the presence of bovine milk proteins in dairy products from other species. However, all these techniques only provide semi-quantitative results and failed to detect small amounts of bovine milk proteins in ovine or caprine products (Muller et al. 2008). Other techniques include analysis of whey proteins, chromatographic techniques (De Frutos et al. 1991) and UV/VIS spectrophotometric detection (Cifuentes 2006) which allow quantitative analysis of milk proteins.

The fraudulent addition of rennet whey solids to dairy products can be determined by the detection and quantification of caseinomacropeptide (CMP), the fragment of κ-CN released by chymosin during milk clotting. It should be absent in normal milk but certain bacterial proteases that degrade κ-CN milk could give rise to the similar degradation products as CMP (Recio et al. 1997). Van Riel & Olieman (1995) successfully separated Chapter 1, 77 the different forms (non-glycosylated forms) of CMP in their studies. It can be concluded that adulterations of dairy products with soya or milk from different species can be detected based on CE patterns.

Muller et al. (2008) demonstrated the use of CE coupled with ESI-MS to determine the adulteration of ovine/caprine milk by bovine milk. This technique was chosen due to electrophoretic clean-up of the milk samples and subsequent identification and quantification of latoglobulins. It was concluded that CE-MS analysis was a rapid, simple and accurate technique to determine and quantify amounts of bovine milk in milk of other species.

Another study on adulteration of dairy products was performed by Guarino et al. (2010) in which LC-MS/MS has been employed to identify and quantify the percentage of sheep‟s milk in cow‟s and goat‟s cheeses. Plasmin was chosen over other proteolytic enzymes for proteolysis due to simple chromatographic separation, mass spectra analysis and simple, fast and widely used protocol. This method appeared to be suitable to monitor adulteration as it allows quantitative detection of the use of sheep‟s milk in goat‟s or cow‟s cheeses.

1.6.2 Peptidomics in Bovine Milk/Bovine Mastitis Due to the importance of peptides in animal physiology, various studies have dealt with this approach using a wide range of samples (cells, tissues or body fluids). In total, a very limited number of peptidomic studies were done to analyze the direct effect of intramammary inflammation on the changes of levels of peptides in bovine milk. However, an application of various techniques of protein/peptide separation coupling with MS (as described above) can be used to study the proteolytic activity in milk samples as proteolysis is increased during mastitis (Verdi et al. 1987).

Apart from plasmin, a principal milk enzyme which hydrolyses β-CN into γ-CN and proteose peptones and to a smaller extent hydrolyses αS1- and αS2-CNs; there are a number of native proteinases in milk derived from somatic cells or leaked out through the damaged somatic cells which also contributes to proteolytic activity in bovine milk (Larsen et al. 2006). PMN which are the main cellular immune components that increase during infection (Azzara & Dimick 1985) contain several enzymes including serine proteases, cathepsin G and elastase (Kirschke & Barrett 1987; Travis 1988; Owen & Campbell 1999). The lysosomes of somatic cells are also thought to contain cysteine proteases (cathepsin B, H, L and S) as well as acid proteases (cathepsin D) (Owen & Campbell 1999). Chapter 1, 78

During mastitis, the increase of somatic cell counts is one of the important disease manifestations. With increasing cell counts, it was estimated that the peptide count in milk was also increased by a factor of 100 (0.03 to 2.71mg/ml) (Lindmark-Mansson et al. 2006). It is also thought that in high SCC milk, proteolysis is caused more by proteases apart from plasmin (Larsen et al. 2004; Kelly et al. 2006). In cheese production, higher SCC contribute to lower yield and poorer quality of cheese (Grandison & Ford 1986), increased moisture content of cheese and the activity of lysosomal enzymes in cheese

(Marino et al. 2005) and less elastic with soft cheddar cheese due to degradation of αS1-CN (Creamer & Olson 1982).

Wedholm et al. (2008) have studied the presence of peptides produced from proteolytic activities of different level of SCC in milk and demonstrated the responsible proteases based on their cleavage sites. Using RP-HPLC to separate the peptides and further characterized by MALDI-TOF MS/MS, peptides that successfully identified were in the range of mass from 1023 to 2000 Da which originated from αS1-, αS2- or β-CN. It was concluded that plasmin, cathepsin B, D and leucocyte elastase were responsible for these proteolytic activities.

Larsen et al. (2010) studied the proteolytic cascade in a model system which used the bacterial lipotechoic acid (LTA) or gram-positive S. aureus to mimic inflammation of the mammary gland. They have also employed RP-HPLC and MALDI-TOF MS/MS in this study and concluded that approximately 20 different peptides were identified originating from αS1- and β-CN. The responsible proteases were plasmin, cathepsin B and D, elastase and amino- and carboxypeptidases.

In summary, by employing various methods of peptidomic analyses, the presence of peptides in mastitic bovine milk can be studied and characterized. These biological peptides may contribute to the discovery of important and novel mastitis biomarkers. Furthermore, identification of these peptides may be applied in the development of therapeutic interventions of bovine mastitis.

1.7 Animal-applied Metabolomics Metabolomics is one of the new technologies that has gained interest and can be considered as important as other “-omics” studies such as genomics, transcriptomics and proteomics. The metabolome is a set of low-molecular weight metabolites that are present in a biological fluid, cell or organisms under different physiological states (Alvarez- Chapter 1, 79

Sanchez et al. 2010). According to Kell (2006), metabolomic analysis is more promising for several reasons. Metabolites appear to be in smaller sizes, thus the sample processing can be reduced as compared to genes and proteins. It is also evident that preparation for metabolic profiling is less expensive than proteomics and more samples can be analysed to provide a good and reliable results and finally metabolomic technology is more generic than any other analysis.

A problem that the metabolomic analysis faced is lack of comprehensive and standardized methods. In order to make sure a generalized detection and identification of metabolites, an integration of various analytical approaches is being made [e.g., gas chromatography combined with mass spectrometry (GC-MS), liquid chromatography combined with mass spectrometry (LC-MS) and nuclear magnetic resonance(NMR)].

The implementation of metabolomics in milk and investigations of the pathophysiology of animal diseases has not been the objective of much research. To date, there is no major study on the metabolomics in bovine mastitis but there are relevant publications.

Metabolomic analysis has been used in milk in order to monitor a whole range of metabolic processes. One classical metabolomic measurement approach is by using chromatography (Martens et al. 2007) to analyse the fatty acid profile in bovine milk sample where it was used to quantify with variation in their chain lengths and unsaturation properties.

Another study by Boudonck et al. (2009) revealed the use of liquid and gas chromatography that successfully identified over 223 metabolites between 10 different milk varieties. Metabolic profiling was shown to possibly provide a distinct biochemical signature in different types of milk and a specific marker could be assigned which could describe specifically for particular milk.

Apart from examining milk metabolomics, the method has been applied to a few investigations of animal models. Metabolomic analyses have been used to study the metabolic changes in canine liver disease (Whitfield et al. 2005). Plasma samples were analysed using LC-MS and metabolites that were identified to increase significantly with liver disease were the taurine conjugates of the bile acids, cholic and chenodeoxycholic acids whilst 16:0-, 18:2- and 18:0-lysophosphatidylcholine were decreased in diseased animals. This analysis was able to distinguish animals with congenital portovascular anomalies from those with acquired syndromes (Moore et al. 2007). Chapter 1, 80

Bertram et al. (2004) have studied splanchnic metabolism in cattle using proton nuclear magnetic resonance (H NMR) from arterial, portal and hepatic plasma. It was concluded that the methyl groups of propionate, 3-β-hydroxybutyrate and lactate were increased in spectral intensities in the H NMR portal plasma samples. As for hepatic plasma samples, due to hepatic uptake of short fatty acids (n-butyrate, propionate, iso-butyrate, valerate and iso-valerate), there was reduced in spectral intensities of these regions. In addition, a hepatic excretion of β-hydroxybutyrate and glucose resulted in increased in spectral intensities of these regions. These spectral differences can be used to describe changes in metabolites which could be supported by biochemical analyses.

In another study by Bertram et al. (2009), they have used the similar approach to investigate propylene glycol toxicosis in a Holstein cow. However, no significant metabolite differences could be detected in the biofluids of the cow which developed toxicosis. It could be due to the vitality of the metabolites or they might be present below the detection limit of NMR. Although there was a decreased in content of isopropanol,isobutyrate, β-hydroxybutyrate, acetate, acetone and acetoacetate shortly after feeding with propylene glycol, it might be due to reduce in fermentation activity in the rumen as well as reduce absorption in ruminal atony. It can be concluded that shallow breathing manifested by the affected animal was due to pulmonary vasoconstriction from the sulphur-containing compounds produced during fermentation of propylene glycol. However, no sulphur-containing compounds were observed in the NMR spectra.

The influence of different lactation stages on physiological and metabolic variables has been studied by Klein et al. (2010) using high resolution nuclear magnetic resonance (NMR) and mass spectrometry (MS). They have identified 44 milk metabolites in milk samples from individual cows during early and late lactation with an emphasis on metabolites related to energy metabolism. There was a large difference in the levels of acetone and β-hydroxybutyrate (BHBA) during early lactation. These two compounds are important biomarkers for subclinical ketotic conditions. The large discrepancies values suggest that different individual animals cope very differently with the metabolic stress during this period and it is highly true for high-producing animals.

In conclusion, the new technology of metabolomics has made it possible for researchers to study the effect of disease progression on the metabolome and could provide new, low molecular weight biomarkers to assess the health condition of animals. However, due to high diversity and variability of metabolome, there are no available analytical platforms that can define metabolites simultaneously (Dettmer et al. 2007). Thus, an efficient Chapter 1, 81 sampling and separation of metabolites should be achieved so that low-abundance metabolites can be detected with the use of small sample volumes. Furthermore, data analysis and visualization tools, libraries, and databases for metabolomics need to be developed so that standardization in data analysis can be achieved.

1.7.1 Metabolites found in bovine milk A number of studies have been performed to evaluate the presence of metabolites in bovine milk with earlier, non-metabolomic methods. Investigations which compared the presence of metabolites between mastitic milk samples and healthy milk samples (assuming low somatic cell counts and negative microbiological culturing) will be discussed further in Chapter 5, Section 5.1.2.

Amongst many metabolites found in bovine milk, fatty acids are one of the important molecules in milk. About 96% of lipid in bovine milk are triacylglycerols and were comprised of a molecule of glycerol and 3 esterified fatty acids (Jensen 1995). Fatty acids on the other hand are composed of 3 different groups based on the number of double bonds present in their chemical structure. They are saturated fatty acids (SFA), monounsaturated fatty acids (MUFA) and polyunsaturated fatty acids (PUFA). MUFA and PUFA form the unsaturated group of FA (UFA) (Soyeurt et al. 2011). Previously, the level of FA in milk was determined by GC after an extraction, followed by saponification and transmethylation but this method was expensive and time consuming. Thus, Soyeurt et al. (2011) have proposed to use mid-infrared spectrometry to quantify the individual content of FA in bovine milk. Ferrand et al. (2011) have also performed a mid-infrared spectrometry to estimate the main fatty acids in bovine milk by applying the genetic algorithm method to improve the quality of estimation. The variation observed in the milk fatty acid composition was determined by several factors such as diet; forage to concentrate ratio (Griinari et al. 1998) and intake of unsaturated fatty acids (Dhiman et al. 1999) as well as seasonal which consequently affected the dietary intake of the cows (Lock and Garnsworthy 2003).

An earlier study on the metabolite levels and enzyme activities in bovine mammary gland has been studied by Waldschimdt and Rilling (1973) at different stages of lactation. Previously, studies have been focussed on the mammary gland of laboratory animals for example rats (Wang 1960; Baldwin & Cheng 1969). This study however determined the level of metabolites in the udders of heifers and cows at peak and near the end of lactation as well as during dry period. Pyruvate, lactate and AMP were not significantly different at Chapter 1, 82 different stages of lactation but quantities of malate, ATP and ADP were higher in cows with high milk yields. Additionally, pyruvate levels showed variations with season whilst the NAD: NADH ratios in mammary cytoplasm showed no significant differences between different groups of cattle. ATP: AMP ratios in mammary gland tissues were highest in lactating animals.

Hollis et al. (1981) studied the presence of vitamin D and its metabolites which include 25- hydroxyvitamin D, 24,25-dihydroxyvitamin D, 25,26-dihydroxyvitamin D and 1,25- dihydroxyvitamin D in both human and bovine milk using chromatographic purification method coupled with ligand-binding assay. It is known that these antirachitic sterols in milk were the same as those in plasma though the concentration in milk is much lower. They are secreted into the milk, bound to plasma and/or cytosol binding proteins (e.g. vitamin D binding protein). Eventually these sterols migrate from the aqueous portion to the fat portion of the milk. Furthermore the levels of anticharitic sterols appear to be linked with dietary intake of these sterols.

With the use of reverse-phase high-performance liquid chromatography (RP-HPLC) Tiemeyer et al. (1984) determined all pyrimidine and purine compounds in milk to relate the nucleic acid metabolites to different rates of protein biosynthesis. Biosynthesis of proteins in the mammary gland is often accompanied by nucleic acids biosynthesis and degradation and consequently a turnover of pyrimidine and purine compounds. In this study, it was established that bovine milk contains nucleosides (uridine, cytidine, pseudouridine) and free bases (hypoxanthine) as well as ortic and uric acids. There are mainly two sources of metabolites in the milk; 1) derived from mammary cell and leukocytes or 2) possibilities of 5‟-nucelotidase activity on the cell surface as described for blood cells. There are also two groups of metabolites based on the influence of milk production; 1) metabolites which are not affected by milk yield (hypoxanthine,uridine, cytidine and pseudoridine) and 2) decrease in amounts with increase in the rate of milk production (orotic acid, uric acid and allantoin). The close correlation between the milk yield and protein secretion showed that high milk production were also accompanied by increase rates of nucleic acid turnover which eventually lead to accumulation of nucleic acid metabolites in both mammary cells and in milk. This study showed rate-limiting steps in the catabolic pathways of pyrimidines and purines with its relationship with protein biosynthesis. Chapter 1, 83

1.8 Aims of study In the past few decades, it has been shown that various studies were employed for protein analysis in bovine milk as well as in bovine mastitis. Although clinical mastitis cases have been widely studied, it is important to take into account the subclinical cases of mastitis for the discovery of important protein biomarkers of mastitis. With the development of new and advanced non-gel approaches (peptidomics and metabolomics), a number of lower molecular weight molecules such as peptides and metabolites can be discovered which are significant in detecting the presence of bovine mastitis. Thus, the overall objective of this study is to apply various proteomic and other non-gel methodologies (peptidomics and metabolomics) to discover significant disease biomarkers in bovine mastitis.

Chapter 2, 84

Chapter II: Acute Phase Proteins in Bovine Mastitis

2.1 Introduction

2.1.1 Bovine Acute Phase Proteins and Acute Phase Reactions As described in Chapter 1, Section 1.4, acute phase proteins (APP) are a group of blood proteins that are synthesized and change in concentration following various stimuli such as infection, inflammation, surgical trauma, stress, neoplastic growth or immunological disorders (Gordon & Koy 1985; Gruys et al. 1999, Murata et al. 2004) which initiate the acute phase reaction (APR) which is also part of non-specific innate immune responses that involves restoration of homeostasis and inhibits the growth of pathogens (Murata et al. 2004; Eckersall & Conner 1988).

Following internal or external challenges, the body reacts by the release of pro- inflammatory cytokines as well as the activation of vascular system and inflammatory cells at the site of microorganism invasion (Gruys et al. 2005). The systemic reaction activated by cytokines caused an induction of hepatic metabolism which leads to the released of APP (Gruys et al.1994). Of the pro-inflammatory cytokines, interleukin-6 (IL-6), interleukin-8 (IL-8), tumour necrosis factor α (TNFα) and interleukin-1β (IL-1β) are known to mediate the synthesis of APP (Nakagawa-Tosa et al. 1995; Alsemgeest et al. 1996; Yoshioka et al. 2002). Some of the APPs are increased during the APR while others, such as albumin, transthyretin (TTR, formerly called prealbumin), retinol binding protein (RBP), cortisol binding globulin, transferrin and albumin, which represent the negative APPs, are decreased (Eckersall & Conner 1988; Gruys et al. 2005). The positive APPs, which are mainly glycoproteins synthesized by hepatocytes, include haptoglobin, C-reactive protein (CRP), SAA, ceruloplasmin, fibrinogen and α 1-acid glycoprotein (α1-AGP) (Murata et al. 2004; Gruys et al. 2005).

Overall, the released of APP during inflammatory condition helps to bind and remove harmful substances following tissue damage, to protect host from further damage and lastly to eliminate pathogenic organisms (Baumann & Gauldie 1994). Chapter 2, 85

2.1.2 Bovine Haptoglobin Haptoglobin (Hp) is a major APP in bovine which strongly binds to haemoglobin, has anti- inflammatory properties and binds to CD11b/CD18 receptor of effector cells (El Ghmati et al. 1996). It also reduces the oxidative damage associated with haemolysis (Yang et al. 2003). Upon binding with haemoglobin (Hb), this stable complex of Hp-Hb is scavenged by phagocytes after recognition via CD 163, the specific surface receptor on macrophages (Schaer et al. 2002).

Bovine plasma Hp was recognized by Liang (1957) and Neuhaus and Sogoian (1961). However, negligible levels of Hp are seen in healthy cows with physiological levels in the plasma being less than 0.1mg/ml (Ritcher 1974). Upon immune stimulation, its circulating level can increases up to 100-fold (Conner et al. 1988; Conner et al. 1989).

Other than binding to haemoglobin, Hp also has an inhibitory effect on granulocyte chemotaxis, phagocytosis and bactericidal activity (Rossbacher et al. 1999). According to Xie et al. (2000), Hp is able to prevent spontaneous maturation of epidermal Langerhans cells or suppress T-cell proliferation (Baseler and Burrell 1983).

Hp is often associated with inflammatory conditions in cows. Increased Hp levels were seen in cattle with mastitis (Spooner & Miller 1971; Conner et al. 1986), liver abscesses and other inflammatory diseases including pyometra, reticulitis, traumatic pericarditis (Panndorf et al. 1976; Makimura & Suzuki, 1982). Moreover, detection of Hp as a highly polymeric complex with Hb was found in cattle with acute inflammation (Morimatsu et al. 1991). APP was not stimulated by viral diseases for example in inflammation by rinderpest virus (Spooner & Miller 1971) or in bovine leukosis (Panndorf et al. 1976). However, Eckersall et al. (1988) found smaller Hp response in some viral infections compared with bacterial infections. Just like any other APP, Hp concentration can be used to determine the severity status of the disease and the extent of tissue damage in affected cows (Murata et al. 2004). Thus, Hp has the potential on being a significantly useful parameter to measure the occurrence and severity of inflammatory responses.

Apart from infection, the level of Hp is raised in cows with fatty liver syndrome (Yamamoto et al. 2000; Katoh 2002), by starvation, from dexamethasone treatment (Yoshino et al. 1993), at parturition (Uchida et al. 1993), and in stressed calves associated Chapter 2, 86 with travelling (Murata & Miyamoto 1993). However, Skinner et al. (1991) found that in cows with normal births, there was no significant increase in Hp indicating that Hp elevation at parturition is due to infection rather than trauma.

2.1.3 Haptoglobin in bovine mastitis Bovine Hp has been demonstrated to be a selective indicator for the acute phase responses in animals. Skinner et al. (1991) demonstrated that Hp can be used to diagnose early bacterial infection, particularly mastitis. The Hp test is also known to be accurate, can be performed quickly and has similar sensitivity as other major bovine APP; the SAA (Boosman et al 1989). Skinner et al. (1991) concluded that the concentration of Hp between 0.2g/litre-0.4g/litre and above was an indication of infection.

Many studies have demonstrated that bovine Hp is one of the important APPs that could be used to diagnose bovine mastitis. It also has been proven that serum Hp levels raised with both experimental and naturally occurring diseases (Conners et al. 1986; Salonen et al. 1996; Hirvonen et al. 1996; Skinners et al. 1991; Hirvonen and Pyorala 1998; Eckersall et al. 2001; Ohtsuka et al. 2001).

Previously, serum bovine Hp was used to investigate its potential in diagnosing bovine mastitis. This is due to the fact that Hp is liver-derived protein (Koj 1974), synthesized by the hepatocytes in the event of acute phase reaction. As described previously, during the activation of acute phase reactions, the concentration of Hp can increase up to 100-fold from its normal level. Furthermore, its level in the normal serum is negligible and thus, the measurement of bovine Hp in serum has been used to analyse the effect of acute phase reaction during inflammatory processes. The increase of bovine serum Hp concentrations during mastitis is supported by various studies (Conner et al. 1986; Salonen et al. 1996; Horadagoda et al. 1999; Ohtsuka et al. 2001).

However it is now known that Hp can be detected in milk during mastitis. The first study that determined the presence of Hp in milk was performed by Eckersall et al. (2001). Measuring Hp level in milk to investigate the presence of intramammary inflammation was suggested to provide a more readily and accessible indication of infection. According to Nielsen et al. (2004), the use of APP in milk to assess the APR following inflammatory processes provides a more specific and sensitive diagnostic method than bacteriology and it is less influenced by the physiological stage of the cow than SCC or EC (Sheldrake et al. 1983; Biggadike et al. 2002). According to Grӧnlund et al. (2003), although increase in Chapter 2, 87 serum Hp occurred during mastitis, it was not specific as increases in serum APP can be due to other acute inflammatory conditions, but measuring Hp in milk provides specificity.

It is first thought that an increased of Hp level in milk during bovine mastitis was due to transfer of these serum proteins synthesized largely by the hepatocytes into the mammary gland due to increased permeability of blood-milk barrier (Eckersall et al. 2001). This is based on the presence of other types of serum proteins such as serum albumin and protease inhibitor α-antitrypsin in the infected quarters as reported by Honkanen-Buzalski et al. (1981) and Honkanen-Buzalski & Sandholm (1981). Grӧnlund et al. (2003) also stated it was not clear whether APP found in the infected milk are locally produced or by an influx of these proteins from blood due to the change in the permeability of the blood-milk barrier.

However, Nielsen et al. (2004) suggested that higher Hp concentrations in the milk from infected quarters could be due to the production of these proteins by the mammary tissues. They also suggested that increased serum Hp concentration in the cows with mastitis was due to the access of Hp synthesized by the mammary tissues to the blood serum. Hiss et al. (2004) confirmed that bovine mammary gland can be considered as another extrahepatical source of Hp based on the observed Hp mRNA in mammary gland tissues (the teat, the cisternal region, the glandular parenchyma). The authors also suggested that the migrated leukocytes might contribute or solely account for the detection of Hp mRNA. Thielen et al. (2005) further demonstrated that Hp mRNA was present in bovine peripheral blood leukocytes and milk somatic cells and confirmed that leukocytes are another cellular source of mammalian Hp mRNA expression.

When comparing the increased level of Hp in milk to that of serum, it was found that the former occurred at 15h after experimentally induced mastitis by S. aureus while the earliest detection of increased serum Hp was at 24h postinfusion. Blood was therefore unlikely to be the source of immediate Hp elevation in milk (Eckersall et al. 2006). The probability of local production of Hp in the mammary gland was further supported by quantitative RT- PCR.

A study by Lai et al. (2009) proposed that Hp levels correlated well with SCC in bovine milk due to the fact that SCC >500 000 cells/ml indicated 10-fold increase in milk Hp concentrations. Thus, neutrophils which were the predominant cell type in a group with Chapter 2, 88

>500 000 cells/ml, are the major source of Hp synthesis. As the conclusion, Lai et al. (2009) suggested four combined pathways for the presence of Hp in milk during mastitis which were: activation and migration of neutrophils by inflammatory cytokines into the infected mammary gland where they secrete cellular Hp; direct entrance of plasma Hp into mammary epithelial cells via a receptor-mediated process and secreted into alveolus; passive diffusion of plasma Hp into udder; endogenous synthesis and expression of Hp by mammary epithelial cells.

Åkerstedt et al. (2007) found the presence of Hp was significantly more common at higher SCC in quarter and composite milk samples but not in bulk tank milk samples due to dilution that took place from quarter to bulk tank level.

2.1.4 Quantification of bovine Hp The determination of bovine serum Hp was first demonstrated by Bremner (1964) but the concentrations were too low to be detected in healthy cattle but local inflammation induced by turpentine injection was able to increase the level of Hp in serum. Various types of assays have been described to determine Hp levels directly or indirectly. There were haemoglobin binding assay (Owen et al. 1960) modified by Jones & Mould (1984) using a microtitre plate format while Skinner et al. (1991) used an automated method based on an RA500 analyser. Immuno assay has also been described, for instance single radial immunodiffusion (Morimatsu et al. 1992). It is known that Hp binds with free Hb with high affinity. Measurement of the inherent Hb peroxidase activity which is protected from acid inactivation by formation of the Hb-Hp complex was used to determine Hp levels in plasma (McNair et al. 1995). Amongst many methods Hb binding assay is the most common method used to determine Hp level in cattle.

Makimura & Suzuki (1982) determined the level of Hp by performing the modified method of Tarukoski 1966 based on the peroxidase activity of haptoglobin-haemoglobin (HpHb) complex. This modified method provided more reliable values even for haemolysed serum samples. Polyacrylamide gel disc electrophoresis (PAGE) employed in this study only provided a qualitative but simple and reliable method to determine Hp.

However, due to bovine Hp being highly polymerised polymorphic protein (Eckersall & Conner 1990; Morimatsu et al. 1991), it can disturb both Hb-binding and immunochemical assays (Salonen et al. 1996) due to its large and heterogeneous molecular sizes. Chapter 2, 89

Depolymerisation of bovine haptoglobin with 2-mercaptoethanol produced homogenous haptoglobin with a high cyanmethaemoglobin (HbCN)-binding capacity.

Hp can also be analysed photometrically based on the principal that cyanomethaemoglobin (HbCN) when bound to Hp was protected from denaturation in an acidic medium (formate buffer, pH 3-7) as described by Elson (1974). Studies by Salonen et al. (1996), Hirvonen et al. (1996) and Hirvonen et al. (1997) used this photometric assay to analyse serum Hp and the absorption of the acidified Hp-HbCN complex was measured at 405nm (Harvey 1976) and 380nm for the sample blank. The haemoglobin binding capacity was calculated in order to determine Hp concentrations.

High performance liquid chromatography (HPLC) can also be used to analyse serum Hp by binding of depolymerised Hp by excess HbCN and separating Hp-HbCN complex and free HbCN by gel filtration on HPLC (Salonen et al. 1996). This method proved not to be influenced by haemolysis in the sample, haemoglobin is converted to HbCN by Drabkin’s solution and the excess did not interfere with the assay thus making it more sensitive and reproducible than that of photometric assay in detecting bovine serum Hp.

Due to the development of monoclonal antibodies specific for bovine Hp (Sheffield et al. 1994; McNair et al. 1995; Godson et al. 1996), antibody-based methods promised the advantages of direct and specific measurement of Hp (McNair et al. 1997). Indeed, Sheffield et al. (1994) used monoclonal antibody against bovine Hp in ELISA to measure bovine Hp level in sera.

Morimatsu et al. (1992) developed a single radial immunodiffusion method (SRID) to overcome problems of measuring bovine serum Hp using Hb-binding assay. Purified bovine Hp from acute phase bovine sera and specific antisera from rabbits were used in this method. Hp was treated with 20mM cysteine or glutathione (GSH) to partially reduce the Hp to smaller molecules or homogenous form.

McNair et al. (1995) introduced a highly sensitive and specific competitive time resolved fluorimetric assay to determine bovine serum Hp. It was shown to enhance the level of detection using a solid phase competitive immunoassays in combination with lanthanide chelate labelling. This method used mouse monoclonal antibody (1D9) specific for bovine Hp which was labelled with lanthanide chelate. It has been shown that this method has a Chapter 2, 90 superior sensitivity when compared with Hp binding assay since there was a direct measure of Hp by specific antibody while the Hp binding assay peroxidase activity may be affected by the presence of haemoglobin, active and inactive forms of Hp (Cerda & Oh 1990) and other serum peroxidase inhibitors such as cysteine and glutathione.

Young et al. (1995) further developed the immunoassay described by Sheffield (1994) using either three formats; (a) a direct binding format; (b) a direct Hb binding format and (c) a competitive inhibition Hp immunoassay. These three formats provide simple, rapid results but with different reproducibility. Furthermore, the reagents used in these immunoassays were inexpensive, non-hazardous and readily available. It was concluded that Format b gave a very good correlation with Hp levels determined by the standard peroxidase activity method with good precision and reproducibility as compared with other two formats.

Immunoassay detected elevated Hp concentrations (62.5%) compared with binding assay (19.5%) although a close relationship (r=0.94) was found between the two methods (McNair et al. 1997).

However the majority of methods available relate to its measurement in serum and cannot be used to measure Hp in milk due to lack of sensitivity of the test and interfering substances present in the milk which can overestimated its concentration in milk (Eckersall et al. 2001). Milk Hp may be quantified using agarose immunodiffusion plates as performed by Eckersall et al. (2001).

Subsequently, several studies have employed the use of ELISA to measure milk Hp (Grӧnlund et al. 2003; Nielsen et al. 2004; Eckersall et al. 2006; Safi et al. 2009; Milica et al. 2010; Larsen et al. 2010; Haghkhah et al. 2010) but Hiss et al. (2004) claimed that the immunodiffusion assay and the sandwich ELISA were not sensitive enough to detect Hp level in milk from healthy cows. They eventually established an ELISA for bovine Hp that is highly sensitive to detect Hp levels in milk from healthy cows throughout lactation and physiologically increased Hp levels (e.g; colostrum) (Hiss et al. 2004).

In contrast, ELISA measurement of bovine milk Hp was said to be unsuitable for routine large-scale analysis (Åkerstedt et al. 2006). They developed a biosensor method which Chapter 2, 91 appeared to be rapid and sensitive based on the interaction between Hp and Hb using surface plasmon resonance (SPR) biosensor technology.

2.1.5 Bovine Serum Amyloid A (SAA) The other major bovine APP is SAA which is an acute phase protein complexed to high density lipoprotein (HDL) as an apolipoprotein in serum (Benditt & Eriksen 1977; Skogen et al. 1979; Husebekk et al. 1988). In order to form a complex with HDL, it has to compete for the binding sites with other apoproteins (Husebekk et al. 1987). However, a study by Husebekk et al. (1988) found that no formation occurred of complexed amyloid A (AA) protein-HDL which suggested removal of SAA from the HDL complexes before degradation to amyloid A in amyloid deposits took place. Serum amyloid A (SAA) is thought to be a precursor protein for AA, a major constituent in bovine amyloid fibrils (Husebekk et al. 1985). However, one of the most important features of SAA is that it can increase more than one hundred-fold in concentration during inflammatory reactions (Horadagoda et al. 1999).

The molecular weight of SAA in its native form is estimated at 18 kDa (Pepys & Baltz 1983). SAA is involved in the transport of cholesterol from dying cells to hepatocytes (Liang & Sipe 1995), has an inhibitory effect on fever (Shainkin-Kestenbaum et al. 1991), on the oxidative burst of neutrophilic granulocytes (Linke et al. 1991) and on in vitro immune response (Aldo-Benson & Benson 1982; Benson & Aldo-Benson 1979). It has a chemotaxic effect on monocyte, polymorphonuclear leukocytes and T cells (Badolato et al. 1994; Xu et al. 1995), induction of calcium mobilization by monocytes (Badolato et al.

1995) and inhibition of platelet activation (Zimlichman et al. 1990)

SAA is synthesized predominantly in the liver with an induction of various pro- inflammatory cytokines such as IL-1, IL-6 and TNF-α (Jensen & Whitehead 1998; Sipe 2000). It is also known to be produced extra-hepatically; by macrophages, endothelial cells and smooth muscle cells (Dibartola & Benson1989), intestinal epithelium (Vreugdenhill et al. 1999) as well as mammary epithelial cells (McDonald et al. 2001).

According to Nazifi et al. (2008), several isotypes of SAA are found in bovine; type 1 and 2 which represent positive acute phase proteins in bovine, while Jacobsen et al. (2004) also successfully determined different types of SAA isoforms in plasma/serum of cows with mastitis. The acute phase response also caused hepatic expression of SAA1 and SAA2 which are collectively known as the A-SAA proteins in a study by Jensen & Whitehead Chapter 2, 92

(1998). Meanwhile, in a range of human tissues which include breast, pancreas and intestine, the mRNA encoding SAA1, SAA2 and SAA 4 proteins has been detected by in- situ hybridization (Urieli-Shoval et al. 1998).

Amongst many isoforms, SAA3 seems to be the predominant SAA isoform in most mammals which expressed the protein extra-hepatically (Ramadori et al. 1985; Meek and Benditt 1986; Benditt and Meek 1989). McDonald et al. (2001) discovered the mammary- associated SAA3 (M-SAA3) isoform with a pI value of 9.6 was present in high levels in colostrum of equine, bovine and ovine. Although it is found in high concentrations in colostrum, M-SAA3 expression in the mammary gland is part of natural physiological process that is not only limited in disease-associated responses. Its presence provides beneficial effects for the newborn as well as for the maintenance of mammary gland.

2.1.6 Serum Amyloid A in Bovine Mastitis SAA as discussed previously can be used as a marker of tissue injury in cows (Horadagoda et al. 1999) and thus is considered to be a potential diagnostic marker due to its low level in normal animals but with a significantly increased concentration during an acute phase response and a later decrease in concentration with the resolution of the disease (Gruys et al. 1994).

SAA response during inflammation/infection is known to be longer in duration when there were co-infections with virus and bacteria rather than in pure viral infections (Ganheim et al. 2003). Its magnitude and duration also depended on the severity of the clinical signs (Ganheim et al. 2003; Heegaard et al. 2000).

The reliability of SAA in comparing between acute and chronic inflammation has been studied by Horadagoda et al. (1999) where it was observed that SAA levels were significantly higher in cows with acute inflammation than in cows with chronic inflammation. It was proved that SAA was the most reliable indicator in differentiating cattle with acute and chronic inflammation and confirmed that the use of APP measurement was better than a haematological test, such as neutrophil count for the presence of infection.

Many studies have been done in order to investigate the potential of SAA as one of the important disease markers in bovine mastitis. Eckersall et al. (2001) first measured the SAA in mastitic milk as well as serum of cows with mastitis. It has been shown that the Chapter 2, 93 measurement of SAA in milk is able to discriminate between mild and moderate mastitic cows compared with SAA in serum. The ability of the SAA in providing sensitive results is due to the local synthesis of SAA in the mammary gland or milk which induces a greater response in moderate mastitis. The presence of different isoform of SAA in mammary gland was confirmed by McDonald et al. (2001) in which M-SAA3was identified in healthy bovine colostrum. There is debate on the naming of the isoform in milk. While it should be named M-SAA3 if it can be definitely identified as this isoform, most assays measure all isoforms and here it will be referred to as M-SAA3 depending on the publication being cited.

As shown by Grӧnlund et al. (2003), SAA levels in serum elevated dramatically during acute mastitis but these results can be due to other acute inflammatory conditions (Conner et al. 1988; Heegaard et al. 2000). Compared with the results from Horadagoda et al. (1999), Grӧnlund et al. (2003) found much lower levels in serum APPs during chronic phase of mastitis than serum APP levels during acute mastitis. Milk serum amyloid A3(M- SAA3) however provided a good marker in chronic subclinical mastitis as it was significantly higher in infected quarters than in pre-infected and healthy udder quarters (Grӧnlund et al. 2003).

To be useful as a mastitis indicator, SAA needs to accumulate in milk during episodes of intramammary infection (Nielsen et al. 2004). Its high level in milk was found to coincide with the onset of clinical signs of acute mastitis as well as increases in SCC in milk (Grӧnlund et al. 2003). However, in some cases, APPs in milk started to be elevated in concentration in milk from clinical mastitic cows even before the SCC started to rise (Hogarth et al. 2002). This is in agreement with Pedersen et al. (2003) where the concentration of M-SAA3 increased prior to increase in SCC in cows induced with S. uberis.

High levels of SAA in infected quarters can be attributed to its local production in mammary gland as well as from leakage of serum proteins across the blood-milk barrier due to increased permeability during mastitis (Nielsen et al. 2004). According to Petersen et al. (2003), it can be assumed that SAA is produced locally in the mammary gland due to the lack correlation between SAA in milk and serum (Eckersall et al. 2001), the expression of serum amyloid protein homologue in the mammary gland (Molenaar et al. 2002) and the discovery of milk specific form of bovine SAA by McDonald et al. (2001). Chapter 2, 94

SAA levels were also found to be increased during endotoxin-induced mastitis in both serum and milk of the cows (Lehtolainen et al. 2004). However, serum SAA was not related to M-SAA3 as SAA seemed to peak earlier in milk than in serum. It was also thought in this study that some SAA that was present in serum may originate from milk due to changes in permeability of blood-milk barrier. However, Jacobsen et al. (2004) disagreed on this by saying that udder-specific SAA was unable to cross the blood-milk barrier and thus be present in the plasma. Lehtolainen et al. (2004) also concluded that M- SAA3 was less able to assess the severity of mastitis because of the slow rise of M-SAA3 concentration during intramammary infection.

In the study by Jacobsen et al. (2004) using E. coli-induced mastitis, they found several SAA isoforms in both plasma and milk of the cows. However, the presence of alkaline pI, tissue-specific SAA3 isoforms in the milk did not appear in any of plasma or serum samples examined confirming the synthesis of SAA in the mammary gland. Furthermore, M-SAA3 levels in milk started to rise earlier at 12-hour post inoculation than SAA levels in plasma (between 12-hour to 24-hour post inoculation). Thus, M-SAA3 became a suitable marker for early mastitis detection as well as an indicator of tissue damage severity.

Larson et al. (2005) showed that bacterial LPS that caused a severe mammary gland infection were able to induce differential expression of M-SAA3 by bovine mammary epithelial cells. It was shown that inflammatory cytokines induced by LPS played an important role in the upregulation of M-SAA3 expression in the mammary gland and initiate a cascade of events during the acute phase response. The A-SAA proteins expressed hepatically also present in the mastitic milk as part of leakage of serum proteins into the mammary gland from increased permeability of the blood-milk barrier.

SAA levels also known to be increased both in serum and milk with repeated challenges of E.coli inoculation in udder quarters. However, serum SAA levels (24-hour post challenge) increased later than the levels of M-SAA3 in milk (12-hour post inoculation) but peaked at the same time in two challenges. They further confirmed the use of APP as mastitis indicator and to assess the severity of inflammation (Soujala et al. 2008).

S. aureus infection often leads to subclinical and chronic mammary gland inflammation Chapter 2, 95 and further promotes the spread of disease to other cows (Weber et al. 2005). Lipotechoic acid (LTA) from gram-positive bacterial cell wall components mimicked a mastitic episode as induced by Gram-positive bacteria (S. aureus) without having the confounding factors of bacterial growth as well as epithelial cell invasion. LTA was shown to be able to stimulate the differential production of SAA by mammary ductal cells with M-SAA3 (pI 9.6) being the predominant SAA isoform produced (Weber et al. 2005).

Another experimental study of S. aureus induced mastitis by Eckersall et al. (2006) supported the finding that increases in M-SAA3 level in milk occurred earlier than increases of SAA in serum. The level of M-SAA3 was also greater in quarters that received 2 infusions of bacteria compared with quarters that only received single infusion of bacteria. Immunocytochemistry results revelead that M-SAA3 is present in the secretory epithelium, gland cistern and at significantly lower levels in the teat canal and it was concluded that the origin of M-SAA3 in milk was within the secretory epithelial cells and cells lining the gland cistern. Less M-SAA3 detected in the teat canal was probably due to infusion of bacteria directly into the gland cistern by-passing the teat canal.

Milica et al. (2010) further demonstrated that the magnitude of both local and systemic APR in S. aureus mastitis was higher if more than 1 udder quarter were affected. Higher magnitude of local tissue response was observed due to different strain of S. aureus (Dingwell et al. 2006), number of S. aureus (Eckersall et al. 2006) and additional exudation of very high plasma SAA that leads to increased in concentration of MAA in milk.

When comparing the reliability of APP measurement in diagnosing bovine mastitis with other mastitis diagnostic methods, APP showed significant correlations with ATP as in a study by Grӧnlund et al. (2005), about 52% of quarters of negative and positive ATP agreed with the results on APP level in udder quarters (ATP-Hp-SAA- group and ATP+Hp+SAA+). However, discrepancy in results was observed between APP level and SCC in affected udder quarters possibly due to different mechanisms of entry into the udder. Leukocytes migrate actively into the mammary gland by means of adhesion molecules while APP is presumed to leak passively from blood into milk from altered membrane permeability or is produced locally by the mammary gland.

O’Mahony et al. (2006) revealed that there was a good correlation between the Chapter 2, 96 concentration of SAA in bovine milk and established cellular indices of mastitis particularly the CMT and SCC. CMT in full and SCC have a strong correlation as both represent the cell concentrations in milk. However, there were some indications that negative correlation happened between the level of milk SAA and SCC in a group of cows when the level of SAA was increased. Nevertheless, this study suggested that milk SAA as useful in detecting quarters with high SCC as serum SAA did not significantly influence the M-SAA3.

Milk APP also showed high correlation with SCC as rise in milk APP concentrations was seen with increasing in SCC thresholds (Kovac et al. 2007). Thus, measurement of milk APP can be used in monitoring udder health as SCC was not suitable to be used alone as a marker to identify infected quarters (Zecconi & Piccini 2002).

Åkerstedt et al. (2007) further confirmed the significant relationships between SAA and SCC at quarter, cow composite and bulk tank level of healthy milk samples. SAA was also detected more frequently than Hp regardless of the types of milk samples (quarter, cow composite and bulk tank) at higher SCC.

The detectable level of M-SAA3 in milk can also be used to predict the protein quality of cow composite milk samples in which samples with detectable SAA level had lower casein number, lower lactose level and higher SCC. Casein number provided a good and specific milk quality parameter especially in cheese-making industry as total protein may include less valuable serum proteins that are known to increase in concentrations during mastitis. Reduced casein number could be contributed by proteolysis while reduced lactose content might be due to the regulation of lactose secretion by specific peptides released during casein hydrolysis by plasmin (Silanikove et al. 2000).

Gerardi et al. (2009) revealed that by using milk a SAA ELISA kit to measure the level of SAA in milk, a significant correlation was found between SCC and M-SAA3 in lower SCC levels from quarter milk samples compared with the level of M-SAA3 in milk determined by serum SAA ELISA kit. Thus, M-SAA3 has been suggested as a physiological marker of subclinical mastitis as it provided statistically better response in distinguishing subclinical from clinical mastitis.

2.1.7 Quantification of bovine SAA In early studies, a variety of SAA assays have been employed but faced with certain Chapter 2, 97 difficulties because most of them require some form of plasma or serum denaturation by heat, acid or alkali so that complete dissociation of SAA from the HDL 3 fraction of lipoprotein occurred (Marhaug et al. 1983; Sipe et al. 1989). This resulted in considerable variation in the reproducibility of the assays.

In one example, Horadagoda et al. (1993) purified amyloid A from acute phase bovine serum and quantified the SAA concentration in serum by an indirect ELISA using rabbit anti-human SAA and horseradish peroxidase conjugated donkey anti-rabbit IgG. However, this type of assay was considered to be laborious, time consuming and depended on the commercial availability of antisera to human amyloid A with cross-reactivity to the bovine protein (Horadagoda et al. 1999). They suggested that by using immunoturbidimetric or latex agglutination immunoassay it was quicker and based on antiserum to bovine SAA. Latex agglutination nephelometric immunoassay has been used to measure human SAA on an automated analyser in less than 6 minutes (Yamada et al. 1993).

McDonald et al. (2001) then developed a monoclonal antibody sandwich immunoassay for SAA that did not require sample denaturation and can be performed in 3 hours. It also has other advantages such as the hybridomas producing the monoclonal antibodies are self- renewing, inexpensive to maintain and yield a highly consistent, easily purified product, the microtitre plates can be coated and stored so that the assay is available within 3 hours, it provides an accurate and reproducible results and lastly requires equipments and skills that normally available in a research or a clinical laboratory.

Later, a commercialized SAA ELISA (Tridelta Development plc, Ireland) was produced based on the assay first developed by McDonald et al. (1991). With that, studies of SAA measurement whether in serum or milk have been done using commercialized SAA ELISA.

In the subsequent chapters in this thesis, measurement of Hp and M-SAA3 in milk has been used to confirm the origin of milk samples from udder quarters with mastitis. The immunoassay for bovine Hp used in previous studies (Eckersall et al. 2006) was no longer available and a novel ELISA was developed for this purpose. While attempts were made to establish an ELISA for M-SAA3, this was not successful and a commercial assay for SAA was used in these investigations. Chapter 2, 98

2.1.8 Aims of study The aims of this chapter were to develop sandwich bovine Hp assay before validating and optimizing this assay to test its reliability to measure Hp. Commercialized SAA assay using Tridelta™ kit was used to measure bovine milk SAA. This type of SAA assay was used due to several problems encountered in developing an in-house bovine milk SAA assay especially in getting a good and acceptable standard SAA curve. After validating and optimizing these assays, they were used to measure APP in milk for the next three chapters to confirm the disease status of the animals.

2.2 Materials and Methods

2.2.1 Reagents

General chemicals and other materials were obtained from Sigma Chem. Co, Poole, UK and Bio-Rad Laboratories Inc. for PAGE unless otherwise stated. Double distilled water was used throughout.

2.2.2 Milk samples

Milk samples were obtained from Cochno Farm, West of Scotland and Vale Veterinary Laboratory, Dorset, UK. Samples from subclinical mastitic cases (n=10) were chosen based on their previous somatic cell count records, no obvious clinical signs as the time of sampling but produced positive California Mastitis Test (CMT) results. Samples from clinical mastitic cases (n=40) were chosen based on their clinical signs and further confirmed by bacteriological culture results which mainly identified major mastitis pathogens such as E. coli, S. aureus and Strep uberis. Samples from healthy animals (n=25) (SCC < 100000 cells/ml) were used in comparison with animals with mastitis cases. Samples were stored at -20°C until required.

2.2.3 California Mastitis Test

The CMT procedure was conducted as discussed in the previous chapter on Section 1.2.2. The CMT reagent dissolved or disrupted the outer cell wall and the nuclear cell wall of any leukocyte and DNA was released from the nuclei. A stringy mass was developed and the amount of gel formation increased with an increase number of leukocytes (Mellenberger 2001). The result was interpreted as presented in Table 1.1 in Chapter 1.

2.2.4 Development of Bovine Haptoglobin ELISA Chapter 2, 99

2.2.4.1 Biotinylation of antibody The biotinylated rabbit anti-bovine Hp was prepared using a biotinylation kit according to the manufacturer’s instructions as described below. Bovine anti-haptoglobin was purchased from Life Diagnostic Inc., USA. This antibody was stored at the concentration of 3.43mg/ml. Different biotinylation of antibody was made up at least for 4 times and the reason will be discussed in the discussion part (Section 2.4).

2.2.4.1.1 Preparation of Elution buffer and Antibody Binding Solution Elution buffer was prepared by adding 50μl of 4M Imidazole stock solution to 950μl PBS. The Antibody Binding Solution (ABS) was prepared by adding 29μl of anti-bovine Hp to 210μl of PBS.

2.2.4.1.2 Antibody Binding The disc provided was put in the bottom of a 1.5ml microcentrifuge tube. The prepared ABS was added to the tube containing the disc. To hydrate the resin, the tube was incubated at room temperature for 30 seconds. The tube was inverted several times to suspend the resin and further 10 minutes incubation at room temperature with gentle rocking motion on a rotating platform was performed (no vortex). It must be noted that the resin must remain suspended during binding and the tube was manually inverted every 2-3 minutes to keep the resin in suspension. The tube was centrifuged at 500 x g for 30 seconds to pellet the resin. The liquid was carefully removed with a pipette to discard. 1mL of PBS was added to the tube and the tube was inverted several times to wash the resin. The tube was centrifuged at 500 x g for 30 seconds to pellet the resin and the liquid was carefully removed with a pipette to discard. To complete washing, 1mL of PBS was added to the tube as before and step was repeated. 300μl of PBS was then added to the tube containing antibody and resin and pipetting the liquid gently up and down to resuspend the resin. A Handee Spin Column was placed in a new 1.5ml microcentrifuge tube. The entire volume of resin slurry was pipetted into the spin column. The column was centrifuged at 500 x g for 30 seconds and the flow-through was discarded before placing the column back into the tube.

2.2.4.1.3 Antibody Biotinylation A bottom plug was applied to the separation column. A No-Weigh NHS-PEO4 Biotin Microtube was punctured with a pipette tip and the tube contents were dissolved by adding 200μl of PBS with gentle pipetting. 188μl of PBS was added to 12μl of biotin reagent Chapter 2, 100 directly to the separation column. PBS was added to the column first before adding the biotinylation reagent. The column was capped with a screw cap and gently mixed by flicking. The column was incubated for 1 hour at room temperature with the column being flicked occasionally during incubation to keep the resin from settling. The bottom cap of the column was removed and it was centrifuged at 500 x g for 30 seconds. The flow- through was discarded and the column was placed back into the same tube. 400μl of PBS was added to the column and further centrifuged at 500 x g for 30 seconds. Again, the flow-through was discarded and the column was placed back into the same tube. The PBS was added as before and the step was repeated three times to wash the column.

2.2.4.1.4 Antibody Elution The column was capped at the bottom and placed in a new 1.5ml microcentrifuge tube. 200μl of Elution buffer was added to the column. The tube was incubated at room temperature for 10 minutes before centrifugation at 500 g for 30 seconds to elute antibody. After elution, some antibody will remain bound to the column. To increase yield of biotinylated antibody, 200μl of Elution buffer was added and incubated for 10 minutes at room temperature and elution repeated. The resin was discarded after used. The biotinylated antibody was stored at 4°C for up to one month. Biotinylated antibodies are generally stable when stored in Elution buffer (0.2M Imidazole in PBS) at 4°C. If biotinylated antibodies were not used within one month, they were stored in single-use aliquots at -20°C.

2.2.4.2 Newly prepared biotinylated antibody To test for the availability of newly prepared biotinylated antibody, serial dilutions of standard bovine Hp (3.43mg/ml, Life Diagnostics Inc., USA) were prepared in 1:100, 1:200, 1:400, 1:800 and 1:1600. Standard bovine Hp at a different dilution was blotted onto a nitrocellulose membrane. 10% of Marvel milk in TBS Tween buffer was used to block the membrane for 60 minutes. The membrane was rinsed with wash buffer and the newly prepared biotinylated antibody at 1:1000 was blotted onto the membrane and further incubated at 60 minutes. 1:1000 of streptavidin peroxidise (Tridelta Development Ltd., Ireland) was blotted after rinsing and incubated for another 60 minutes. The membrane was rinsed and substrate (Opti-4CNTM Substrate Kit, USA) was added to develop bands. Chapter 2, 101

2.2.4.3 Assay Procedure The assay conditions were optimised as described in section 2.2.4 so that the best performance from the assay is achieved. The following protocol described the finalised conditions.

2.2.4.3.1 Coating The Life Diagnostic rabbit anti-bovine haptoglobin as described earlier was diluted to a final concentration of 1 µg/ml in 0.05M NaHCO3, pH 9.6. 100 µl was dispensed into each well of a Nunc-Maxisorp 96 MicroWell™ plate (Nunc International, Rochester, NY) and incubated at 4°C overnight.

2.2.4.3.2 Washing Wells were washed using 0.02M Tris-HCl pH 7.4 with 0.05% Tween-20 by adding 250µl per well for 4 times.

2.2.4.3.3 Blocking Unoccupied binding sites were blocked by adding 250µl of 10% (w/v) Marvel milk protein in wash buffer and incubated at 37ºC for 60 minutes.

2.2.4.3.4 Washing Wells were washed again as above.

2.2.4.3.5 Dilution and addition of Standard Bovine Haptoglobin and milk samples Standard bovine haptoglobin (0.845mg//ml, Life Diagnostics Inc., USA) was prepared at 0.25µg/ml in washing buffer and double diluted until concentration reached 1.95ng/ml. Milk samples were also diluted at 1:10000 or 1:1000 in washing buffer. 100µl of each standard bovine haptoglobin and milk samples were added into duplicate wells and incubated at 37 °C for 60 minutes.

2.2.4.3.6 Addition of biotinylated antibody After washing, diluted biotinylated rabbit anti-bovine haptoglobin at 1:1000 in washing buffer was added at 100µl per well and incubated at room temperature for 60 minutes.

2.2.4.3.7 Addition of secondary antibody After washing, diluted Streptavidin-peroxidase at 1:1000 in washing buffer was added at 100µl per well and further incubated at room temperature for 60 minutes. Chapter 2, 102

2.2.4.3.8 Addition of substrate After washing, Tetra methyl benzidine (TMB) peroxidise substrate was prepared according to manufacturer instruction (KPL, Inc., USA). The substrate was added at 100µl per well and developed for 5-10 minutes. The peroxidase-enzyme reaction was stopped by adding 100µl per well of 1M hydrochloric acid.

2.2.4.3.9 Absorbance reading The absorbance was read at 450 nm using Fluostar OPTIMA platereader (BMG Labtech Ltd.) and the results analysed and calculated using the associated FLUOstar OPTIMA Software V1.32 R2 using a 4 parameter-fit standard curve plotted on a linear scale.

2.2.5 Haptoglobin Assay Optimisation

During assay development a number of parameters were altered to come to the optimised assay. A number of these variations are given below.

2.2.5.1 Antibody concentrations The concentration of primary antibody (anti-rabbit bovine Hp) was varied to achieve the highest optical density (OD) value for the maximum standard while producing a steep standard curve, with low OD for blanks. Other variables were kept constant with varying antibody concentrations of 1µg/ml, 5µg/ml and 25µg/ml. Though the curve for 1μg/ml of antibody gave the lower curve (Fig 2-2) it was chosen for the optimised assay due to its lower background reading.

2.2.5.2 Varying batch of biotinylated antibodies With each batch of antibody prepared, different dilutions of antibody were tested so that a straight standard curve with low OD values for blanks were achieved.

2.2.5.3 Biotinylated antibody concentrations The concentration of biotinylated antibody (anti-rabbit bovine Hp) was varied to achieve aims as previously described. Other variables were kept constant with varying antibody concentrations of 1:250 and 1:1000. Figure 2-4 shows the Hp standard curve for the first batch of biotinylated anti-bHp at 1:250 and 1:1000, the latter was chosen for the optimised assay due to a lower background. Figure 2-5 shows the comparison between batch 1 and batch 2 of the biotinylated antisera at 1:1000. With batch 3 and batch 4 of the biotinylated antibody a dilution of 1:500 was optimal. Chapter 2, 103

2.2.5.4 Optimisation of Standard and Sample Dilution The standard bovine Hp was diluted at different concentrations and serially diluted to achieve standard curves over different ranges, so that a steep standard curve over was produced. Figure 2-6 shows a typical standard curve. Milk samples were diluted at different concentrations until their reading were on the linear section (0-1000ng/ml) of the standard curve. It shows that samples of milk with elevated Hp gave parallel dilution curves with the standard curve giving an indication that the assay is measuring haptoglobin.

2.2.6 Haptoglobin Assay Validation

2.2.6.1 Precision For intra-assay precision, the coefficient of variance was determined as the mean of coefficient of variance (CV) for of duplicate results. Inter-assay precision was determined where CVs were calculated from the mean and standard variation of the same milk samples (control samples) over 10 times during the use of the assay.

2.2.6.2 Accuracy Linearity of dilution of 3 milk samples with varying Hp concentrations was assessed for accuracy of the assay.

2.2.6.3 Limit of detection The limit of detection was determined as the lowest Hp concentration which could be distinguished from a zero value, and was taken as the mean +3 standard deviations of 4 replicates of blank samples.

2.2.6.4 Specificity

The specificity of the assay was based on the commercial anti-bovine Hp anti-serum assessed by Ouchterlony immunodiffusion in agarose gel.

2.2.7 Serum Amyloid A Assay

The concentration of SAA in milk samples were quantified using a commercial Multispecies SAA ELISA kit (PhaseTM Range, Tridelta Development Limited) according to the manufacturer’s instructions. Chapter 2, 104

2.2.7.1 Reagents General chemicals and other materials were obtained from Sigma Chem. Co., (Poole UK) while Tridelta Development Ltd supplied the Phase™ Range SAA ELISA kit.

2.2.7.2 Reagent Preparation All buffers should be prepared beforehand according to the instruction manual. These include 1X diluents buffer, 1X wash buffer, 1X biotinylated anti-SAA conjugate and 1X Streptavidin-peroxidase.

The SAA calibrator provided in the kit was reconstituted by adding distilled water to the vial and vortexed vigorously for 3-4 times for 3-5 seconds to dissolve the pellet completely. Top calibrator was prepared first before serially diluted.

Quality controls from both with highest and lowest SAA values were diluted at 1:500 using 1X diluent buffer.

All milk samples were diluted in 1X diluents buffer before analysis.

2.2.7.3 SAA Assay 50 µl of diluted milk samples, diluted SAA calibrators and quality controls were added to duplicate wells. 50µl of 1X biotinylated antibody conjugate were also added to each well and incubated at 37ºC for 60 minutes. After incubation, the plate was washed 4 times using 1X wash buffer before tapping the plate dry on absorbent paper. 100µl of 1X Streptavidin- peroxidase was added to each of the wells and further incubated at room temperature in the dark for 30 minutes. The plate was washed as above. 100 µl of TMB Substrate were added before incubated at room temperature in the dark for 30 minutes. Finally, 50µl of Stop solution was added and the absorbance was read at 450nm.

2.2.8 SAA Assay Validation

The specificity and accuracy of the SAA ELISA was based on the data supplied with the commercial ELISA (Tridelta Development Ltd., Ireland).

2.2.8.1 SAA Assay Precision For intra assay precision, the CV was calculated from the mean CV of duplicate samples (n=42). Inter-assay precision was determined where CVs were calculated from the mean and standard variation of the same milk samples (control samples) over 10 times during the use of the assay Chapter 2, 105

2.2.8.2 Limit of Dectection The limit of detection was determined as the lowest Hp concentration which could be distinguished from a zero value, and was taken as the mean +3 standard deviations of 15 replicates of blank samples.

2.2.9 Clinical Validation

Milk samples were collected from two different locations; the Cochno Farm, University of Glasgow, Clydebank, UK and Vale Veterinary Group Devon, UK. Milk samples acquired were grouped into healthy which comprised of 25 milk samples (Cochno Farm), subclinical mastitis group which comprised of 10 milk samples (Cochno farm) and clinical mastitis group which were comprised of 40 milk samples (Vale Veterinary Group). Both healthy and subclinical mastitic milk samples were chosen based on their monthly SCC recording data whilst cows with clinical mastitis were chosen based on the general clinical signs of mastitis and further confirmed by bacteriological culture.

2.2.10 Data Analysis

Statistical analysis was carried out using Predictive Analytics Software (PASW) Statistics (SPSS) version 18 (IBM Corporation, NY, US).

2.3 Results 2.3.1 California Mastitis Test

All subclinical mastitic milk samples (n=10) were subjected to CMT and all results revealed positive CMT reaction (2+ or 3+).

2.3.2 Immuno dot blot of newly prepared biotinylated antibody

The newly prepared biotinylated antibody was tested for cross reaction as shown in Figure 2-1. The immuno dots developed with the greatest intensity at lowest dilution (1:100) of standard bovine Hp and decreased as the dilution of Hp increased.

2.3.3 Haptoglobin Assay Optimisation

2.3.3.1 Coating Antibody concentrations Chapter 2, 106

During optimisation, varying concentrations of coating antibodies were assessed as shown in Figure 2-2. A final concentration of 1µg/ml of coating antibody was chosen due to its reasonably steep curve and covered a wide range of OD values whilst giving lower blank readings than other different dilutions.

2.3.3.2 Biotinylated antibody concentrations

The biotinylated antibody was optimised as the previous step with different dilutions of 1:250 and 1:1000 using two different types of antibodies as shown in Figure 2-3. The final concentration of 1:1000 of biotinylated antibody was chosen as it resulted in a steep curve, provided a good range of OD values with lower blank readings than other different dilutions.

2.3.3.3 Varying batch of biotinylated antibodies

For the Batch 1 of prepared biotinylated antibodies, several dilutions were tested (1:1000 and 1:250). The chosen concentration was 1:1000 of as previously mentioned. An example of standard curve is shown in Figure 2-4.

For the Batch 2 of prepared biotinylated antibody using anti-rabbit bovine Hp, the 1:1000 dilution still give the best standard curve as shown in Figure 2-4.

But for the Batch 3 and 4 of prepared biotinylated antibody using anti-rabbit bovine Hp, the 1:1000 dilution did not produce a very reasonable standard curve thus a range of different dilutions were tested (1:250, 1:500 and 1:1000). However, a good standard curve was not achieved.

Chapter 2, 107

Figure 2-1: Immuno dot blot of biotinylated anti-bovine Hp reacting with bovine Hp at dilutions of 1:100, 1:200, 1:400, 1:800 and 1:1600.

Chapter 2, 108

Average OD unit 0.7 0.6 0.5 0.4 1ug/ml 5ug/ml 0.3 25ug/ml 0.2 0.1 0 30 300 Hp concentration (ng/ml)

Figure 2-2: Optimisation of coating antibodies concentrations using different concentrations for the developed ELISA. The effect of anti-rabbit bovine Hp concentrations at 1µg/ml, 5µg/ml and 25µg/ml on bovine Hp ELISA. The 1µg/ml concentration gave lower blanks and a steeper curve than others. Chapter 2, 109

1 0.9 0.8 0.7 0.6 0.5 1: 250 biotinylated 0.4 antibody 0.3 0.2 0.1 1:1000 biotinylated antibody 0 30 300

Figure 2-3: Optimisation of biotinylated antibody concentrations for the developed ELISA. The effect of the biotinylated antibodies at 1250 and 1:1000 on bovine Hp ELISA. The biotinylated antibody at 1:1000 provided lower blanks and a steeper curve than others. Chapter 2, 110

1.4

1.2

0.8 Batch 1 biotinylated antibody 0.6 Batch 2 biotinylated antibody 0.4

0.2

0 1.5 15 150

Figure 2-4: Standard curve obtained from the optimised first and second batch of biotinylated antibodies (3.43mg/ml) diluted at 1:1000.

Chapter 2, 111

2.3.3.4 Optimisation of standard and sample dilution

A working standard curve is needed as part of assay optimisation. This was achieved by diluting the standard bovine Hp (0.845mg/ml) in assay buffer to 0.25µg/ml and serially diluting the standards to achieve concentrations between 1.95-250ng/ml. An example of working standard curve is given on Figure 2-5. Milk samples were diluted accordingly based on their Hp content until the readings were on the linear portion of the standard curve.

2.3.4 Haptoglobin Assay Validation

2.3.4.1 Precision

The intra assay CV for the bovine Hp assay was 2.5% which was calculated as the mean CV of duplicate samples (n=36). Table 2-1 showed the values, means and the calculated CVs for the Hp ELISA assays which were 29.9% and 47.1% for mean milk Hp of 21 and 3.3μg/ml respectively.

2.3.4.2 Accuracy

Serial dilution of three milk samples as shown in Figure 2-6 was used to assess the assay accuracy. The three samples show linearity then plateau out at low dilution and are parallel to the standard curve. Diluted sample concentrations were taken from the linear portion of the curve as it is the most precise and accurate area of the curve.

2.3.4.3 Limit of detection

The minimum detection limit significantly different from the blank samples was taken as the Hp concentration at 3 standard deviations (SD) from the mean of blank sample (n=4). This resulted in a limit of detection of Hp in milk of 0.37ng/ml.

Chapter 2, 112

1.4

1.2

0.8

0.6

0.4

0.2

0 1.5 15 150

Figure 2-5: Standard curve obtained from optimised standard bovine Hp dilution. Standard bovineHp was diluted from 0.845mg/ml to 0.25µg/ml, then by 1:2 dilution. Chapter 2, 113

Sample ID Hp Conc. (μg/ml) Plate Sample ID Hp Conc. (μg/ml) Plate 92 15.4 1 41 6.5 1 92 25.2 2 41 5.5 2 92 15.2 3 41 3.3 3 92 19.4 4 41 3.9 4 92 11.8 5 41 2.5 5 92 21.3 6 41 2 6 92 32 7 41 2 7 92 29.3 8 41 2.9 8 92 20.1 9 41 2.3 9 10

Mean 21.16 Mean 3.31 SD 6.33 SD 1.56 % CV 29.9 % CV 47.1

Table 2-1: Inter-assay precision results obtained for ELISA with milk samples. The CV for Hp concentration of 21.2μg/ml was 29.9% and the CV for Hp concentration of 3.31μg/ml was 47.1%. Chapter 2, 114

2.5

2 Sample 1

1.5 Sample 2 Sample 3 1 Standard curve 0.5

0 0 50 100 150 200 250 300

Figure 2-6: Linearity of 3 bovine milk samples after serial dilution. Example of standard curve included for comparison.

Chapter 2, 115

2.3.4.4 Specificity

The specificity of the assay was based on the commercial anti-bovine Hp anti-serum and was assessed by Ouchterlony immunodiffusion in agarose gel (Figure 2-7).

2.3.5 SAA Assay Validation

The specificity and accuracy of the SAA ELISA was based on the data supplied with the commercial ELISA (Tridelta Development Ltd., Ireland). In use the precision and limit of detection of the SAA ELISA for assessment of the protein in milk were as follows.

2.3.5.1 Precision

The intra assay CV for the SAA assay was 6.8% calculated from the mean CV of duplicate samples (n=42). Results were from too few assays to calculate the inter assay just for this investigation but when included in a series of ELISAs run in the laboratory (n=35) the CVs were 46% at a concentration of 7.2 + 3.3 ng/ml and 30% at a concentration of 40.6 + 12.1 ng/ml. These values are for post dilution of samples and are equivalent to 1.44μg/ml and 8.12μg/ml in the undiluted milk samples respectively. While the CVs were high, they were deemed to be satisfactory in view of the large increased being measured between healthy and diseased samples.

2.3.5.2 Limit of detection

The limit of detection of the SAA was 4ng/ml being the value determined as 3 SD from the mean of replicate blank samples (n=15).

2.3.6 Clinical Validation

The descriptive statistics of all variables in each group are presented in Table 2-2 while the distribution of SAA, Hp concentrations and SCC measurement in all samples are presented in Figure 2-8, 2-9 and 2-10. For healthy group, the range of SAA concentration was between 0.03-2.96μg/ml with a median of 0.36μg/ml, mean of 0.54μg/ml and standard deviation of 0.61μg/ml. The range of Hp concentration fell Chapter 2, 116 between 0.07-31.95μg/ml with the average of 9.17μg/ml with a median of 3.52μg/ml and standard deviation of 9.88μg/ml. The range of SCC measurement was between 12 x 103 to 81 x 103 cells./ml with a median of 43 x 103 cells/ml, mean of 43 x 103 cells/ml and standard deviation of 19 x 103 cells/ml on at least one SCC recording one month before the sampling took place.

For subclinical mastitic group, the range of SAA concentration was between 0.18- 14.05μg/ml with a median of 1.44μg/ml, mean of 3.19μg/ml and standard deviation of 4.27μg/ml. The range of Hp concentration fell between 0.31-14.91μg/ml with a median of 7.41μg/ml, average of 7.49μg/ml and standard deviation of 5.13μg/ml. The range of SCC measurement was between 67 x 103 to 1224 x 103 cells/ml with a median of 268 c 103 cells/ml with the average of 486 x 103 cells/ml and standard deviation of 447 x 103 cells/ml on the SCC recording one month before the sampling took place.

For clinical mastitic group, the range of SAA concentration was between 1.78- 1780.5μg/ml with a median of 63.55μg/ml, average of 422.14μg/ml and standard deviation of 649.08μg/ml. The range of Hp concentration fell between 1.95- 413.85μg/ml with a median of 86.52μg/ml, average of 99.36μg/ml and standard deviation of 90.16μg/ml. The cultured bacterial pathogens were amongst the common causal of bovine mastitis such as E. coli, S. aureus, S. uberis and S. dysgalactiae (Table 2-3)

As the data was not normally distributed, the Spearman’s correlation was performed to analyze the correlation between these variables (Table 2-4). The Spearman’s correlation between SAA and Hp concentration in all groups of milk samples was 0.653. The Spearman’s correlation between SAA and SCC in all groups of milk samples was 0.438 whilst the Spearman’s correlation between Hp and SCC in all groups of milk samples was 0.08.

Mann-Whitney’s test was used to analyze for significant differences between diseased groups on SAA, Hp and SCC measurements (Table 2-5). Healthy and clinical mastitic groups showed significantly difference in the SAA concentration (p<0.05) with the z value of -6.737and healthy and subclinical mastitic groups also showed significant difference in SAA concentration (p<0.05) with the z value of -2.41. Clinical mastitic and subclinical mastitic groups had significant differences in SAA concentration (p<0.05) with the z value of -4.573. Chapter 2, 117

Healthy and clinical mastitic groups showed significantly difference in the Hp concentration (p<0.05) with the z value of -5.437 whilst the Hp concentration between healthy and subclinical mastitic groups was not significantly different (p>0.05) with the z value of -0.087. Clinical mastitic and subclinical mastitic groups however had significant difference in Hp concentration (p<0.05) with the z value of -4.131.

Finally, healthy and clinical mastitic groups showed significantly difference in the SCC measurements (p<0.05) with the z value of -4.439. Chapter 2, 118

Figure 2-7: Ouchterlony immunodiffusion gel of anti-bovine Hp. The centred well was anti-bovine Hp, well 1 was purified bovine Hp, well 2 was low Hp bovine serum, well 3 was low Hp bovine serum, well 4 was high Hp bovine serum, well 5 was high Hp bovine serum and well 6 was high Hp bovine serum. A single line of immunoprecipitate was observed showing that the antiserum was specific for bovine Hp and that there were no non-specific reactions with other proteins in the serum samples.

Chapter 2, 119

Group Std. N Minimum Maximum Mean Median Deviation Variance Healthy SAA 25 .03 2.96 .5435 .3600 .60944 .371 Hp 22 .07 31.95 9.1665 3.5150 9.88022 97.619 SCC 25 12000 80500 42800.00 43000.00 19812.664 3.925E8 Valid 22

N Subclinical SAA 10 .18 14.05 3.1870 1.4350 4.27359 18.264 Hp 9 .31 14.91 7.4848 7.4120 5.13332 26.351 SCC 10 67000 1224000 485900.00 286000.00 446707.573 1.995E11 Valid 9

N Clinical SAA 40 1.78 1780.50 422.1374 63.5480 649.08256 421308.170 Hp 40 1.95 413.85 99.3570 86.5200 90.16136 8129.070 SCC 0 Valid 0

Table 2-2: The descriptive statistics for the Hp concentration (μg/ml), SAA concentration (μg/ml) and SCC of milk samples taken from healthy, subclinical mastitic and clinical mastitic cows. The values for mean, maximum and minimum values, median, standard deviation and variance of each Hp, SAA and SCC for each group of milk samples were presented.

Chapter 2, 120

Sample IDs Bacteriological culture 475 E. coli 76 E. coli 1111 E. coli Star Strep. uberis 1510 Strep. dysgalactiae 814 E. coli 438 S. aureus 940 Coryne. bovis 74 NA 669 Strep. uberis 544 Strep. uberis 708 Strep. uberis 200 Strep. uberis 185 Strep. uberis 14 S. aureus 692 E. coli 1061 NA 697 E. coli 519 E. coli 31 Strep. uberis 980 S. aureus 453 E. coli 496 E. coli 1222 E. coli 828 NA 49 Strep. dysgalactiae 4493 Strep .uberis 669 Strep. uberis 708 S. aureus 474 E. coli 346 S .aureus 7 Strep. uberis 230 Strep. uberis 1512 Strep. dysgalactiae 715 S. aureus 452 E. coli 932 E. coli 770 Strep. uberis 656 Strep. uberis 339 Strep. uberis Table 2-3: The bacteria causing mastitis cultured from the clinical mastitic milk samples. NA stands for not available.

Chapter 2, 121

Figure 2-8: The box-plot showed the distribution of SAA levels in bovine milk samples between healthy, subclinical mastitis and clinical mastitis groups. The median of SAA concentration in healthy group was 0.36μg/ml whilst the maximum and minimum values were 2.96 and 0.03μg/ml respectively. The median of SAA concentration in subclinical mastitic group was 1.435μg/ml whilst the maximum and minimum values were 14.05 and 0.18μg/ml respectively. The median of SAA concentration in clinical mastitic group was 63.55μgml whilst the maximum and minimum values were 1781 and 1.78μg/ml respectively.

Chapter 2, 122

Figure 2-9: The box-plot showed the distribution of Hp levels in bovine milk samples between healthy, subclinical mastitis and clinical mastitis groups. The median of Hp concentration in healthy group was 3.52μg/ml whilst the maximum and minimum values were 31.95 and 0.07μg/ml respectively. The median of Hp concentration in subclinical mastitic group was 7.41μg/ml whilst the maximum and minimum values were 14.91 and 0.31μg/ml respectively. The median of Hp concentration in clinical mastitic group was 86.52μg/ml whilst the maximum and minimum values were 413.85 and 1.95μg/ml respectively.

Chapter 2, 123

Figure 2-10: The box-plot showed the distribution of SCC values in bovine milk samples between healthy and subclinical mastitis groups. The median of SCC measurement in healthy group was 43 x 103cells/ml whilst the maximum and minimum values were 80.5 x 103 and 12 x 103 cells/ml respectively. The SCC measurement in subclinical mastitic group was 286 x 103 cells/ml whilst the maximum and minimum values were 1224 x 103 and 67 x 103 cells/ml respectively

Chapter 2, 124

Correlations SAA Hp Spearman's rho SAA Correlation 1.000 .653** Coefficient Sig. (2-tailed) . .000 N 75 71 Hp Correlation .653** 1.000 Coefficient Sig. (2-tailed) .000 . N 71 71 **. Correlation is significant at the 0.01 level (2-tailed).

Correlations SAA SCC Spearman's rho SAA Correlation 1.000 .438** Coefficient Sig. (2-tailed) . .008 N 75 35 SCC Correlation .438** 1.000 Coefficient Sig. (2-tailed) .008 . N 35 35 **. Correlation is significant at the 0.01 level (2-tailed).

Correlations SCC Hp Spearman's rho SCC Correlation 1.000 .080 Coefficient Sig. (2-tailed) . .670 N 35 31 Hp Correlation .080 1.000 Coefficient Sig. (2-tailed) .670 . N 31 71

Table 2-4: Spearman’s correlation between SAA concentration, Hp concentration and SCC measurements in milk samples. The Spearman’s correlation between SAA and Hp was 0.653 whilst the Spearman’s correlation between SAA and SCC was 0.438. The Spearman’s correlation between SCC and Hp was 0.08.

Chapter 2, 125

Healthy vs. Healthy vs. Clinical mastitis Clinical mastitis Subclinical vs. Subclinical mastitis mastitis SAA Conc. z= -6.737 z= -2.41 z= -4.573 (μg/ml) p=0.00 p= 0.00 p= 0.00 Hp Conc. (μg/ml) z= -5.437 z= -0.087 z= -4.131 p= 0.00 p= 0.931 p= 0.00 SCC (x 10-3/cells) NA z= -4.439 NA p= 0.00 Table 2-5: Mann-Whitney’s Test was performed to test for differences in SAA concentration and Hp concentration as well as SCC measurements between healthy, subclinical mastitic and clinical mastitic group.

Chapter 2, 126

2.4 Discussion Hp and SAA have been studied in this chapter to evaluate their effectiveness in diagnosing bovine mastitis prior to use their use to confirm the status of samples used in the following chapters. After optimising and validating the developed assays for Hp and SAA, their concentrations in different groups of milk samples were then compared statistically. Correlation between SCC values and APP concentrations were also analysed statistically.

For the development of bovine Hp ELISA, several stages were optimised to make sure the assays were reproducible. A final concentration of 1μg/ml of 3.43mg/ml rabbit anti- bovine Hp was chosen whilst 1:1000 or 1:500 (depending on batch) of biotinylated antibody was used to achieve an optimised assay. Both gave a reasonably steep standard curve and covered a wide range of OD values with acceptable low background. Each time a new biotinylated antibody was prepared, a series of optimisation on the varying batch of biotinylated antibodies were done. Different batch of biotinylated antibodies at different dilutions (1:1000 or 1:500) were used due to the large number of milk samples assayed throughout this project. The amount of biotinylated antibody was often less than 0.5ml after its preparation.

The Hp assay was validated by determining the intra-assay CVs of 36 duplicates milk samples and inter-assay CVs of the internal controls. Inter-assay CVs were within the range of 29.9-47.1% with the highest value occurring for a control with lowest level of Hp in milk. In general, CV values for biochemical tests should be lower than 10%, but in the case of immunoassays, the value of 20-30% are generally acceptable, especially if the concentrations of the analyte are low (FDA, 2001). In this study, even though the CV value was above 20%, the large changes in Hp concentration meant that the assay could be used to identify samples with high concentrations of this APP in milk samples.

The PhaseTM Range SAA Assay developed by Tridelta Development Limited, was used to detect the M-SAA3 in the milk samples. It had acceptable precision with an intra assay CV of 6.8% in 42 duplicate samples and a limit of detection adequate to measure M-SAA3 in the healthy milk samples.

Chapter 2, 127

2.4.1 Clinical Validation: Assessment of APP level in healthy, subclinical and clinical mastitic milk samples The final stage of validation of the Hp assay and demonstration of the SAA assay was by clinical validation on assays of the APPs in different groups of milk samples. Some of these samples were used in the following chapters. Hp and SAA assays as described in previous section were performed to assess any differences and correlations between these APP levels in three different groups of milk samples.

2.4.2 Hp level in three different groups of milk samples Based on the quantification of Hp levels in this study, Hp was detected in milk samples regardless of its disease status. Statistically, it was found that the Hp concentration in healthy milk samples (n=22) was significantly different (p<0.05) from clinical mastitic milk samples. This was in agreement with many publications (Eckersall et al. 2001; Gronlund et al. 2003) on the ability of Hp as a disease marker in bovine mastitis. Hp level also showed significantly higher in concentrations (p<0.05) in clinical mastitic milk samples in comparison with subclinical mastitic milk samples. Thus, Hp may be able to discriminate between clinical and subclinical mastitic animals as reported by others (Gronlund et al. 2003; Ohtsuka et al. 2001). However, Hp levels in the milk examined here produced no significant difference between healthy and subclinical mastitic milk samples (p>0.05). This is in an agreement with Gronlund et al. (2003) in which they only observed significantly higher level of SAA in infected quarters than pre-infection and healthy quarters but not Hp. The reason might be due to different types of cytokines produced which lead to variation in the secretion and synthesis of APP (Horadagoda et al 1999) in milk samples with different disease status. Furthermore, during subclinical mastitis, Hp production might not be as great as that produced during clinical mastitis. The other possible reason is might be due to the immunoassay which is not sensitive enough to detect a slight elevation of Hp concentrations in sublinical mastitic group. Other factors such as pathogen virulence and host response of the cows can play a role in the production of APP in infected quarters (Gronlund et al 2004). Nevertheless the Hp ELISA developed in the project was able to identify milk samples from cows with clinical mastitis and its use will be demonstrated in subsequent chapters.

Chapter 2, 128

2.4.3 SAA level in three different groups of milk samples With the commercial ELISA to measure SAA levels in this study, this APP was detected in all milk samples regardless of its disease status. This ELISA was sensitive enough to detect a low amount of SAA in healthy animals.

Statistically, SAA levels were found to be significantly different (p<0.05) between healthy and clinical mastitic milk samples. The high concentration of SAA found in all clinically infected milk samples further confirmed the role of SAA as a biomarker in bovine mastitis. SAA levels also showed significantly higher concentrations in clinical mastitic milk samples compared with subclinically infected milk samples (p<0.05). Furthermore, SAA can also be employed to discriminate between clinical and subclinical mastitis and milk SAA concentrations can also be used to differentiate between healthy and subclinical mastitis as it showed statistically significant difference in milk SAA levels between those two groups (p<0.05). These results are in agreement with several studies which have shown that SAA (or M-SAA3) are elevated in milk during mastitis (Eckersall et al 2001; Grönlund et al 2003; Nielsen et al 2004) and in particular those that show that milk-SAA3 is able to differentiate healthy from subclinical mastitis samples (Grönlund et al 2003; Grönlund et al 2005). Thus use of the SAA ELISA was validated to monitor the concentration of this APP in milk for the investigations in the following chapters.

2.4.4 Correlation between SAA and Hp levels in milk samples As the APP and SCC had been measured in multiple samples it was of interest to determine the correlations between the analytes. In general, SAA level showed a strong and positive correlation (r= 0.653) with the level of Hp in clinical mastitic group. This finding confirmed that SAA and Hp can be used as good mastitis indicators in which they both present in a very low amount in healthy milk samples but can increase significantly after induction of acute phase response. This finding also supported the fact that the measurement of APP during intramammary infection can be done by using milk samples rather than serum/plasma due to the ease of sampling and sample preparation. It also showed that the ELISA used to measure these major bovine APP is reliable and sensitive enough to detect its concentration in milk samples.

Chapter 2, 129

2.4.5 Correlation between SCC and APP levels in milk samples Milk samples of healthy and subclinically infected cows were chosen on the basis of SCC measurement. Thus, it is also important to assess the relationship observed between SCC values and APP measurements. SCC has been a common method for mastitis diagnosis and is still being widely used in most of commercial dairy farms as part of disease monitoring system.

According to Hillerton (1999), the limit of SCC for a healthy quarter should be around 100 000 cells/ml and the SCC for composite milk should not exceed 100 000 cells/ml (Laevens et al 1997). In this current study, the average of SCC in the healthy group is 43 x 103 cells/ml. Statistiscally, the SCC measured between healthy and subclinical mastitic milk samples were significantly different (p<0.05). It proved that SCC is still reliably trusted to diagnose subclinical mastitis as changes of milk composition microscopically are the only signs of subclinical mastitis (Radostits et al 1994). The SCC was not measured in the clinical group as measurement can be misleading in clinical mastitis due to the interference of milk protein clots distorting the cell counting results.

The correlation between SCC and APP in was done and showed a positive but not very strong relationship. The correlation between SCC and SAA (r=0.438) gave a medium relationship whilst the correlation between SCC and Hp (r=0.08) gave a very small relationship. The interpretation was based on Cohen (1988).

Previous studies showed a strong and positive relationship between SCC and Hp (Thielen et al 2005; Lai et al 2009). They claimed that leukocytes of milk somatic cell act as one of the cellular source of mammary Hp mRNA expression and also that these circulating bovine leukocytes might be able to synthesize the Hp and contribute towards the measurement of Hp in milk (Thielen et al 2005). It was also supported by Lai et al (2009) that somatic cells were responsible for the increase of milk Hp as Hp levels measured using ELISA were positively correlated with the SCC and also Hp mRNA levels in cells of high SCC groups were significantly elevated than in cells from normal group.

Although the relationship between Hp and SCC was very low, it should be noted that this correlation was calculated based on all milk samples regardless of the type of milk samples (control or infected). Furthermore, various cut-off values for SCC have been Chapter 2, 130 suggested depending on the conventions used, countries, breed and type of milk samples (Mattila et al 1986; Holdaway et al 1996).

As a summary, the ELISA developed for quantification of Hp in milk samples was a reliable and sensitive enough assay to investigate the Hp levels in milk samples of different groups being able to differentiate clinical mastitic samples from subclinical and healthy samples. Commercial SAA ELISA also proved to be able to analyse SAA concentration in all milk samples studied. The measurement of APP levels in diagnosing bovine mastitis appears to be a useful tool in monitoring udder health for its ability to discriminate the healthy milk samples from infected milk samples. However, at present SCC should also be used in detecting the presence of subclinical mastitis in a herd of cows, although various factors should be evaluated (e.g; stages of lactation, milk production level, stress, breed) before any decision-making, especially on therapeutic measures. Furthermore, SCC and APP measurements should be used together as part of udder health inspection in a herd of dairy cows. For the investigations in subsequent chapters, the ELISAs for M-SAA3 and milk Hp along with SCC and bacteriology were used to provide definitive characterisation of milk samples as being from healthy dairy cows or from udder quarters with subclinical or clinical mastitis

Chapter 3, 131

Chapter III: Proteomics in Bovine Mastitis

3.1 Introduction Proteomics has come forward as a rapidly maturing essential tool in the omics age following the huge success of genomic studies. As mentioned previously, proteomics is a study that involved in separation and identification of a set of proteins existing in a cell, organ or organism (Wang et al. 2007). Studying proteins has been a subject of interest for many researchers as any cell responses to stimuli or stress are manifested by the alteration of protein expression levels (Lippolis & Reinhardt 2010). Protein information is also valuable as protein acts as structural building blocks, conveys information, controlling many chemical reactions as well as being involved in antimicrobial defence mechanisms.

Proteomics can also be defined as a large-scale study of protein expression, protein-protein interactions or post-translational modifications (Cravatt et al. 2007; Lippolis & Reinhardt 2010). Unlike other methodologies, proteomics is able to analyze thousands of proteins in a single study and thus one can demonstrate how cells dynamically change upon stimulation or changes in the environment. In particular, proteomics stands out in its ability for a high resolution and simultaneous analysis of the composition of milk. (O’Donnell et al. 2004).

Previous studies on the milk proteome have utilized many different approaches of proteomic analyses. Details on each study have been discussed earlier in chapter 1. By using various proteomic methods such as 2D gel electrophoresis and MALDI analyses (Anderson et al. 1982; Galvani et al. 2000,2001; Baeker et al. 2002), 2D and microsequencing (Murakami et al. 1998; Yamada et al. 2002) as well as LC-MS/MS (Palmer et al. 2006), a complete milk proteome which also included low abundance milk proteins has been achieved. In this section, the technical aspects of proteomic analyses will be discussed further.

3.1.1 Sample Collection and Preparation

The protein level in any biological fluid is influenced by many factors such as normal metabolism, circadian rhythms of protein production, degradation and environmental conditions (Hale et al. 2003). Upon sampling, protein modifications may also occur which include proteolysis, protein aggregation and chemical modifications. Thus, to minimize variation, sample collection and storage must be standardized. Chapter 3, 132

However, most biological fluids are dominated by relatively few proteins for example in blood, albumin and the immunoglobulins (Igs) predominate (Anderson & Anderson 2002). Thus, removal of these proteins is advantageous to retrieve other lower abundance proteins and to reduce sample complexity. For example, the use of columns containing Cibacron blue (Hale et al. 2003) and anti-albumin antibodies (Steel et al. 2003) or albumin-specific peptides (Pingali et al. 1996) can be applied to remove albumin.

Despite the sensitivity of the mass spectrometers used in proteomics study, several factors concerning the complexity of the sample must be analyzed. These include the number of protein in a proteome, the diversity of protein expression, the degree of protein solubility and the copy number of protein classes to assess the changes (Reinhardt & Lippolis 2008; Graham et al. 2005).

Thus, various fractionation protocols were introduced to reduce the complexity of the mixtures for a more complete identification of proteins. This can be achieved via subcellular fractionation, enrichment strategies, chromatography or gel electrophoresis (Stasyk & Huber 2004). The subcellular fractionation provides important information on the nuclear membrane proteome; for example the isolation of MFGM from bovine milk. With enrichment methods, antibodies or metal ions can be used to isolate the protein of interest from the cell or pathogens. Size exclusion filtration may be applied to separate peptides subsequently sequenced by mass spectrometry (Hunt et al. 1992; Lippolis et al. 2002). Chromatography and gel electrophoresis will be discussed next in this section.

3.1.2 Protein separation

Protein separation is the first main technology in proteomic study and two most popular methods are two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) (O’Farrell 1975; Görg et al. 1998; Rabilloud 2002) and the gel-free procedure; liquid phase separation (Isaaq 2001; Isaaq et al. 2002).

2D-PAGE has been utilized for many years and was first established in the mid 70’s (O’Farrell 1975; Klose 1975). It is able to resolve thousands of proteins and the procedure consists of two steps; separation of proteins based on their isoelectric points (pI) and finally separation based on their molecular weight (Mr). Basically, this method combines isoelectric focusing (IEF) and sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) (O’Donnell et al. 2004; Hiendleder et al. 2005). The IEF gels are available Chapter 3, 133 in broad or narrow pH ranges to achieve different levels of resolution (Lippolis & Reinhardt 2008).

Following gel electrophoresis, a spot pattern on a gel is generated and each spot represents a different protein or its modification form (Heindleder et al. 2005). These protein spots can be cut out and further analyzed by mass spectrometry (Lippolis & Reinhardt 2008).

According to O’Donnell et al. (2004) the advantage of using 2D-PAGE includes higher resolution and ability to separate proteins of similar relative molecular mass (Mr) and pI. However, major drawbacks of this method limit its usage such as an inability to separate single polypeptide chains of masses greater than 150 kDa or less than 8 kDa, limited dynamic range and loading capacity of the immobilized pH gradient (IPG) strip while it is unable to resolve proteins with an extremely basic (pI >12) or acidic (pI < 3) in nature.

To overcome the irreproducibility problem of 2DE, difference gel electrophoresis (DiGE) was developed to minimize the effects of gel-to-gel variation by labelling two samples with two different fluorescent dyes before running them on the same gel (Unlu et al. 1997). DiGE is extremely sensitive as it is able to detect over a >10 000-fold protein abundance range (Viswanathan et al. 2006; Lilley et al. 2002).

In DiGE, equal amount of protein samples are differentially labelled using synthetic N- hydroxysuccinimidyl (NHS) ester derivatives of the cyanine dyes Cy3 and Cy5 before being subjected to 2DE separation and finally scanned to produce two super-imposable images (Timms & Cramer 2008). In order to increase the accuracy of the results, standard can be included and labelled with a third fluorophore through internal standardization (Hiendleder et al. 2005).

Other protein separation methods are based on liquid phase separation such as capillary electrophoresis (CE) and liquid chromatography (LC). They are known to have greater sensitivity, provide a wide dynamic range, are easily automated and can be faster than 2D- PAGE (O’Donnell et al. 2004). A number of separation mechanisms can be applied such as size exclusion, reversed-phase (RP) and ion exchange so that strongly acidic/basic proteins can be analysed. Peptides are separated into fractions based on their hydrophobicity (RP-HPLC) or based on their charge state (strong cation exchange chromatography) (Lippolis & Reinhardt 2008).

A rather new approach to provide better separation is multidimensional protein identification (Mud-PIT) where two or more LC techniques (e.g. ion exchange chromatography and RP HPLC) are used with on-line MS (Washburn et al. 2001). Chapter 3, 134

3.1.3 Protein characterization

Protein characterization is mainly accomplished by mass spectrometers. Basically, all current mass spectrometers can detect and identify peptides in the femtomole (10-15) to attomole (10-18) range (Moyer et al. 2003). After initial protein digestion with trypsin, peptides must be ionized upon entering the mass spectrometer. The two most frequently used ionization methods are electrospray ionization (ESI) (Fenn et al. 1989) and MALDI (Tanaka et al. 1988; Karas & Hillenkamp 1988).

ESI allows direct connection between LC and MS due to the volatility of the HPLC solutions. MALDI requires mixing of the digested peptides with a UV-absorbing molecule to form a crystalline structure. A laser will strike this structure for sublimation of the matrix and ionization before releasing the associated peptides (Lippolis & Reinhardt 2008).

After ionization, the peptide mass is determined by mass analyser. Peptide mass is usually expressed as a ratio of mass divided by the charge of the peptide (m/z). Mass spectrometers are completed by detectors which can range from simpler quadrupole (Q), time-of-flight (TOF) to more complex Fourier transform ion-cyclotron resonance (FTICR) analysers (O’Donnell et al. 2004).

Information gained from mass spectrometers includes mass and sequence of the proteins. MALDI-TOF is used primarily for detecting only protein masses while quadruopole-time of flight tandem mass spectrometry (Q-TOF MS/MS) can be used to provide sequence information (Roepstorff & Fohlman 1984). Post-source decay (PSD) is another technique used to sequence proteins (Spengler et al. 1992).

The next step is protein identification by matching the masses of digested tryptic petides with known masses from a non-redundant protein database (Beavis & Fenyo 2000). This technique is simply achieved by peptide mass fingerprinting (PMF). Peptide fingerprinting is often associated with 2D-PAGE and protein identification is based on the measurement of multiple peptides from one particular protein (Lippolis & Reinhardt 2008). The second method to identify proteins is by using tandem mass spectrometers when PMF data are inconclusive (O’Donnell et al. 2004; Lippolis & Reinhardt 2008).

Generally, proteomics experiments can be either direct identification or characterization of a certain known proteins of interest or indirect/shotgun approach. The ultimate goal of Chapter 3, 135 shotgun approach is to identify all proteins present in a sample. It can be divided into survey or expression proteomics (Lippolis & Reinhardt 2010). In survey proteomics, it gives an overview of types of proteins and grouped them based on their biological process, cellular compartment or molecular function (Lippolis & Reinhardt 2010) while expression proteomics involved with relative quantification where two or more samples are compared with one served as a reference (Bantscheff et al. 2007; Lippolis & Reinhardt 2008; Ong & Mann 2005).

3.1.4 Quantitative proteomics

The next stage of proteomics is quantification for estimates of differential protein expression. DiGE is one of the techniques which represent a real advance in protein quantification by 2DE electrophoresis as described above. Fluorescence characteristics of the CyDyes are used to determine the relative abundance of proteins from two different samples (Wu et al. 2009).

The other two commonly used quantitative techniques use molecular tags to label peptides via an interaction with a free thiol or amine group. The first molecular tags is called isotope-coded affinity tags (ICAT) that uses isotopic labelling of cysteine residues by light or heavy tag containing biotinylated reagents (Lippolis & Reinhardt 2008; Gygi et al. 1999). One sample is labelled with the light tag and the other is labelled with the heavy tag. They are combined and enzymatically cleaved before further analyzed by LC-MS/MS. A newer tag approach called iTRAQ has been commercially available since 2004 and uses a set of 4 tags that possess the same molecular weight but different fragmentation points. It permits relative quantification of many samples (up to 8 samples) at one time (Choe et al. 2007).

Other less commonly used methods to obtain metabolic information on bacterial growth are stable isotope from amino acids in cell culture (SILAC) (Ong et al. 2002; Couté et al. 2007) and surface-enhanced laser desorption ionization-time of flight (SELDI-TOF) to identify enzymes that change expression in response to the presence of substrates such as lactose, glucose and galactose (Gagnaire et al. 2009).

3.1.5 Aims of study

The aims of this study were to determine the changes of milk proteome during the presence of intramammary inflammation (subclinical and clinical bovine mastitis). For comparison, the milk proteome of normal milk samples as controls were also studied using 1DE, 2DE, Chapter 3, 136

DiGE and MALDI-MS analyses. The discovery of new biomarkers especially in subclinical mastitis was also an important objective of this study.

3.2 Materials and Methods

3.2.1 Reagents General chemicals and other materials were obtained from Bio-Rad Laboratories Inc. unless otherwise stated. Double distilled water was used throughout.

3.2.2 Milk samples Milk samples used in this study were of subclinical mastitic cases from Cochno Farm of School of Veterinary Medicine, University of Glasgow, UK (n=5) and clinical mastitis cases from Vale Veterinary Laboratory, Devon, UK (n=12). Bacteriological culture was performed for clinical mastitic milk samples in the Bacteriology Laboratory of Vale Veterinary Laboratory to determine the bacterial cause of bovine mastitis. Meanwhile, control milk samples (n=12) came from healthy cows from the Cochno Farm of School of Veterinary Medicine, University of Glasgow. The healthy milk samples were confirmed by somatic cell counts (<100 000 cells/ml) (Kromker et al. 2001) obtained from the Cochno Farm of School of Veterinary Medicine, University of Glasgow, UK. Prior to proteomic analyses, SAA and Hp assays (Chapter 2) were performed to determine the concentration of both acute phase proteins in each milk sample as confirmation of the health status. Samples were stored at -20°C prior to analysis. California Mastitis Test (CMT) was carried out on the subclinical mastitic milk samples at the Cochno Farm.

3.2.3 California Mastitis Test The California Mastitis Test (CMT) was performed as described in the previous chapter on Section 2.2.3.

3.2.4 Removal of high abundance milk proteins

3.2.4.1 Rennet Coagulation Casein, the most abundant milk protein can be coagulated with the enzyme rennin found in rennet (an extract from the stomach of calves) to split a specific bond in the amino acid chain of the κ-casein macromolecule and form para-κ-casein and a glycol-macropeptide (Leonil & Molle 1991). Thus, the removal of casein from milk can be performed by rennet coagulation to allow the investigation of low abundance proteins. A total of 10μl of 15% (w/v) of calcium chloride was added into 1ml of milk sample in order to improve coagulation before adding the rennet. The sample was incubated at room temperature for 1 Chapter 3, 137 hour before pre-heated at 30ºC using a water bath for 20 minutes. An aliquot of 5 µl of 7.5mg/ml rennet (Sigma-Aldrich, Dorset, UK) was added into 1ml milk sample and incubated at 30ºC for 30 minutes. Finally, the sample was centrifuged at 21 000 x g for 10 minutes and supernatant/whey was kept frozen until further analysis.

3.2.4.2 Acid Precipitation Acid precipitation was used to remove major milk proteins such as caseins as another approach to investigate the properties of noncasein milk proteins (Murakami et al. 1998; Baeker et al. 2002, Yamada et al. 2004). A 0.5ml aliquot of milk sample was used and its pH was altered to pH4 using 1M acid hydrochloric (HCl). After pH adjustment, milk sample was centrifuged at 21 000 g for 10 minutes to remove precipitates. Supernatant/whey was kept frozen until further analysis.

3.2.4.3 Ammonium sulphate (NH4SO4) Fractionation Ammonium sulphate fractionation was employed as another method to remove major milk proteins such as caseins. It was thought that this method was able to reduce any potential for acid induced protein hydrolysis (Hogarth et al. 2004). Saturated NH4SO4 was prepared by dissolving NH4SO4 in ice cold distilled water till no more dissolved. NH4SO4 (100%) was added into milk sample to make up 35% of NH4SO4. The sample was incubated for 1 hour in an ice bath before being centrifuged at 21 000 g for 10 minutes. Supernatant/whey was kept frozen until further analysis.

3.2.5 One-dimensional gel electrophoresis Milk samples used for 1D GE were 5 from individual subclinical mastitic and another 6 from individual healthy cases obtained from Cochno Farm, Glasgow, UK.

3.2.5.1 Reagent Preparation and Stock Solutions All stock solutions and buffers were prepared according to Mini-PROTEAN® 3 Cell Instruction Manual. 12% PAGE gels were also prepared according to the manual (see method in Appendix 1, Chapter 3).

3.2.5.2 Sample Preparation Prior to 1DE analysis, the protein content was measured using a BioRad Bradford Assay (see Appendix 2, Chapter 3). Milk samples were diluted to give a sample at 2-3mg/ml and mixed with SDS Reducing Buffer (10% (w/v) SDS containing 0.5M Tris-HCl, pH 6.8, 2.5ml glycerol and 0.05% (v/v) bromophenol blue; Bio-Rad Laboratories, Inc. USA) at a ratio of 1:1 and heated at 95°C for 4 minutes. Chapter 3, 138

3.2.5.3 Running the gel electrophoresis After the assembly of the electrophoresis module, the sample was loaded into wells with a pipette using gel loading tips. Power was applied to the Mini-Protean® 3 cell and electrophoresis was run at 200 volts for 40 minutes.

3.2.5.4 Gel staining and Image Acquisition After electrophoresis, the gels were stained using colloidal Coomassie blue G dye according to manufacturers’ instructions (Sigma-Aldrich, Dorset, UK) and de-stained in 10% (w/v) acetic acid in 25% (v/v) methanol overnight. Gel images were acquired using Umax Power Look III.

3.2.5.5 In-gel trypsin digestion of Coomassie-stained proteins Chosen gel bands from 1D gel electrophoresis were excised and placed in 1.5ml Eppendorf tubes. These bands were then cut into several pieces using pipette tips. Gel pieces were washed for 45 minutes in 500µl of 100mM ammonium bicarbonate and the wash was discarded. Gel pieces were further washed in 50% acetonitrile/100mM ammonium bicarbonate for another 45 minutes. The wash was again discarded. For reduction, 150µl of 100mM ammonium bicarbonate and 10µl of 45mM DTT were added and incubated at 60ºC for 30 minutes. For alkylation, 10µl of 100mM iodoacetamide was added and incubated in the dark for 30 minutes. Solvent was discarded and gel pieces were washed in 500µl 0f 50% (v/v) acetonitrile/100mM ammonium bicarbonate with shaking for 1 hour. The wash was discarded and 50µl of 100% (v/v) acetonitrile was added to shrink the gel pieces. After 10 minutes, solvent was removed and gel pieces were dried completely in a vacuum centrifuge for 30 minutes. The dried gel pieces were left at 4ºC for 15 minutes before adding 0.2µg/µl sequencing grade modified Porcine Trypsin (Promega #V111) in 25mM ammonium bicarbonate to rehydrate each gel piece. Additional trypsin solution was added if trypsin was completely absorbed so that the gel band became fully rehydrated. Finally about 20µl of 25mM ammonium bicarbonate was added over the gel pieces which were incubated overnight to allow protein digestion at 37ºC. 30µl of 100% acetonitrile was then added to the gel pieces and incubated for 20 minutes. Each solution in each tube was transferred to separate wells in a 96-well plate and dried in the Speed Vac. About 20 μl of 1% (v/v) formic acid was added into each Eppendorf tubes containing gel pieces and incubate for another 20 minutes. Each solution was transferred to the corresponding wells. These two steps were repeated for better recovery. About 40µl of acetonitrile was added and further incubated for 10 minutes before the solution was transferred to the corresponding wells. This step was also repeated for one time. The plate was checked at least after 90 minutes to make sure the sample had been dried up. Chapter 3, 139

3.2.5.6 Sample Plating The MALDI plate should be cleaned and rinsed before use. A saturated solution of cyano- 4-hydroxy cinnamic acid (CHCA) in 50% (v/v) acetonitrile and 0.1% (v/v) trifluoroacetic acid (TFA) was prepared in 1.5ml Eppendorf tube and covered with aluminium foil. The tube was vortexed for 5 minutes before being centrifuged for 5 minutes. The saturated solution of CHCA was diluted at 1:10 in 50% acetonitrile and 0.1% TFA. 10µl of 50% acetonitrile and 0.1% TFA was loaded into the plate and 1µl of dried peptide was dissolved into the same spot. 1µl of 90% (v/v) CHCA solution (matrix solution) was added to the peptide on the plate and was mixed well by pipetting up and down for 5-7 times. After sample loading, the plate was left to dry before being analysed by the MS which was supervised by Kamburapola Jayawardena of the Sir Henry Wellcome, Functional Genomic Facilities, University of Glasgow, UK.

3.2.6 Two-dimensional gel electrophoresis Milk samples used for 2D GE were pooled together from six of subclinical mastitic milk samples to make up pooled subclinical mastitic sample and another six from healthy cases were pooled together to make up pooled healthy samples.

3.2.6.1 Sample preparation Each supernatant/whey sample from rennet coagulation was precipitated with acetone overnight for 500μg protein. To a sample volume containing 500 μg, a 10 times volume of ice cold acetone was added and incubated at -20°C overnight. The protein solution was then centrifuged at 14 000 x g for 10 minutes. Supernatant was removed after centrifugation. 1ml of 80% acetone was used for washing before suspending the pellet in 80% acetone. The dissolved pellet was centrifuged at 14 000 x g for another 10 minutes. Supernatant was discarded before dissolving the pellet in 1ml of lysis buffer (see Appendix 3, Chapter 3). To make sure the pellet did not stick to the centrifuge tube’s wall, another brief centrifugation was performed. For the pellet to fully dissolve, the tube was put on the vibrating platform.

3.2.6.2 Isoelectric Focusing A protein concentration assay was run on each sample so that an equal amount (100μg) was present before adding 3-10 IPG buffer (0.5% v/v), DTT 0.001 g/μl (0.02% w/v) and rehydration buffer (6M urea, 2M thiourea, 4% CHAPS and 0.02% (w/v) bromophenol blue) for 100μg. Sample was centrifuged at maximum speed for 5 minutes. About 450 μl of prepared sample was applied on pH3-10, 24cm IPG strip. Isoelectric focusing was run on the IPGphor system using this protocol: Chapter 3, 140

50μA per strip at 20°C

S1: Step-and-hold 30V 12 hours

S2: Step-and-hold 300V 1 hour

S3: Gradient 1000V 2 hours

S4: Gradient 8000V 2 hours

S5: Step-and-hold 8000V 8 hours

S6: Step-and-hold 1000V 24 hours

S7: Step-and-hold 1000V 24 hours

After the focusing finished, the strip was put in a plastic tube with the gel part facing upward and stored at -20°C before running the SDS-PAGE gel electrophoresis.

3.2.6.3 SDS-PAGE Gel Electrophoresis Both IEF Equilibration Buffers (EB) I and II (see Appendix 4 and 5, Chapter 3) were prepared prior to gel electrophoresis. 10ml of EB I was added onto the strip and mixed for 15 minutes. The strip was then washed with running buffer and then 10ml of EB II was added onto the strip and mixed for another 15 minutes. The strip was washed once again before 1ml of running buffer was added on top of the gel (large format SDS PAGE gel). The strip was placed horizontally on top of the precast Ettan DALT® gels (26 x 20cm) and added 1ml of 0.5% agarose (heated at 105°C at least 5 hours prior to gel electrophoresis) was placed on top of the gel. The prepared gel was placed in the tank and the remaining running buffer was added in the tank. The electrophoresis was run overnight at higher and constant power.

3.2.6.4 Gel staining and Image Acquisition After electrophoresis, the gel was fixed for 1-2 hours using 10% methanol and 7.5% acetic acid. Gel was then stained with freshly prepared Colloidal dye solution and put on the rocker overnight. Gel images were acquired using a Typhoon™75 multicolour fluorescence and phosphor image scanner.

3.2.7 Difference Gel Electrophoresis Milk samples used for DiGE comprised 12 clinical mastitic and 12 healthy milk samples which were then pooled together to make up a pooled clinical mastitic milk samples and Chapter 3, 141 pooled healthy milk samples. Concentration of Hp and M-SAA3 were determined as in Section 2.2.3 and 2.2.6.

3.2.7.1 Sample Preparation Cold acetone (-20ºC) at four times sample volumes was added to protein (i.e: 250µl of each sample into 1000µl of cold acetone). This mixture was incubated overnight and it was then centrifuged for 10 minutes at 13 000 x g. The supernatant was decanted and the pellet was washed with 4 times volume of 80% acetone. Centrifugation was performed again and supernatant was decanted. Brief centrifugation was done and residual supernatant was removed with pipette tip. The dry pellet was air dried for approximately 5 minutes at room temperature but was not overdried the pellet. The pellet was resuspended in 10µl of DiGE lysis buffer (see Appendix 6, Chapter 3) with protease inhibitor cocktail with brief spinning. The milk samples were solubilised by probe sonification for 3 times brief cycles (1-2 second each cycle). Between sonification, the sample was cooled on ice for 1 minute. Foaming and heating during sonifcation was avoided. After sonification, the samples were incubated for 10 minutes at room temperature with gentle mixing. Lastly, to remove insoluble material, the samples were centrifuged for 10 minutes at 13 000 x g. The supernatant was transferred to a fresh microcentrifuge tube.

3.2.7.2 CyDye Labelling 1µl of CyDye was added into 10µl of sample and incubated in the dark for 30 minutes. To stop the reaction, 1µl of 10mM lysine was added and further incubated in the dark for 10 minutes. The pooled milk samples from healthy cows were labelled with Cy3 and the pooled milk samples from cows with clinical mastitis were labelled with Cy5.

3.2.7.3 Isoelectric focussing The samples (1 labelled Cy3 and 1 labelled Cy5; ~24μl of each) were mixed together and 500µl of rehydration buffer was added into the sample mixture. Before centrifugation, the sample mixture was left at room temperature for 10 minutes. 450µl of sample mixed in rehydration buffer was pipetted along the length of the strip holder, avoiding any formation of bubbles. The IPG strip was put onto the strip holder before layering mineral oil over the top of the strip, just enough to cover the strip. The lid was placed on the strips which were placed on the IPGphor system (GE Healthcare UK) and isoelectric focussing was run for between 70-80000V hours on the IPGphor as described in Section 3.2.6.2

3.2.7.4 SDS-PAGE After isoelectric focussing, the strips were taken out of the holders and put into the plastic tubes. 10ml of SDS Equilibration Buffer I (see Appendix 4, Chapter 3) were placed onto Chapter 3, 142 the strips and the plastic tubes was put on a large flat rocker desk for 15 minutes. The buffer was poured off and SDS Equilibration Buffer II (see Appendix 5, Chapter 3) was added and was rocked gently for another 15 minutes. The buffer was poured off and the strip was placed on the gel.1ml of 0.5% agarose was added on top of the gel (precast Ettan DALT® gels; 26 x 20cm) before putting the gel in the tank. Electrophoresis was run overnight at 1-2V/gel in the running buffer.

3.2.7.5 Gel Scanning DiGE gels were fixed before scanning on a 3 Laser Typhoon 9400 scanner. A preliminary scan was performed with one DiGE gel at a low resolution (1000 microns). Spot saturation was checked on each of the resulting images, and the photomultiplier tube voltage adjusted so that the most intense protein spots were not saturated (see DiGE manual GE Healthcare). Gels can be scanned in pairs. Each high resolution (100 microns) scan takes > 1 hr to complete but it was found that DiGE gels can be stored wrapped in the fridge overnight without noticeable loss of resolution. Once satisfactory images were obtained, gels were discarded and plates cleaned. After staining and scanning, the gel was wrapped and stored at 4°C until spot picking.

3.2.7.6 Image manipulation DiGE images were cropped using Imagequant software on the Typhoon scanner. The cropping removed all edges of the gel image (gel spacers, IPG strip and dye front). Images were then loaded into DeCyder for processing.

3.3 Results

3.3.1 SCC measurement and acute phase proteins concentration In order to confirm the disease status of milk samples used in this study, SCC measurement and acute phase proteins (SAA and Hp) concentrations were determined. Table 3-1 showed the SCC measurement and CMT scores in all control (n=6) and subclinical mastitic (n=5) milk samples used in this study for 1D and 2D gel electrophoresis. In control milk samples, the range of SCC measurement in control milk samples (n=6) was between 6 and 27 x 103 cells/ml with an average of 18 x 103 cells/ml and median of 19 x 103cells/ml. The range of SCC measurement for subclinical mastitic milk samples (n=5) was between 211 and 4472 x 103 cells/ml with an average of 1355 x 103 cells/ml and median of 389 x 103 cells/ml. The CMT scores for all subclinical mastitic Chapter 3, 143 milk samples were 2+ and all control milk samples gave negative results on CMT score. Table 3-2 showed the SCC measurement concentrations in all control milk samples (n=12) and APP concentrations as well as bacteriological culture results in clinically infected milk samples used in this study for DiGE. In control milk samples, the range of SCC measurement in control milk samples (n=6) was between 6 and 105 x 103 cells/ml with an average of 40 x 103 cells/ml and median of 27 x 103 cells/ml. The range of Hp concentration was between 560 and 16803.62 μg/ml with an average of 5532μg/ml and median of 40234μg/ml. The range of SAA concentrations in clinically infected milk samples was between 16.57 and 10635μg/ml with an average of 1041μg/ml and median of 107.7μg/ml. From the results of bacteriological culture of clinical mastitic milk samples, bacteria that had been successfully identified include S. uberis and E. coli although in some milk samples, no growth of bacteria was identified.

3.3.2 Comparison between rennet coagulation, acid precipitation and salt precipitation (ammonium sulphate) to remove casein After each of casein precipitation methods, whey samples from normal milk were run on 1D gel electrophoresis for protein separation. For comparison, whole milk without any casein precipitation was run on the 1D gel electrophoresis as a control. Based on Figure 3- 1, Lane 2, protein bands of casein were found between 25-37 kDa by comparing to Mr standard (Lane 1) showing the presence of major milk protein in the whole bovine milk. As for Lane 3, 4 and 5 where various methods of casein precipitation were applied, protein bands of casein were completely absent from the 1D gel. Furthermore, both rennet coagulation (Lane 3) and acid precipitation (Lane 4) showed increased intensity of protein bands between 10-15 kDa of molecular weight. These protein bands are likely to be β- lactoglobulin (15 kDa) and α-lactalbumin (10 kDa). However, more low intensity of higher molecular weight protein bands were seen between 50-250 kDa on both rennet precipitation (Lane 3) and acid precipitation (Lane 4).The intensity of those high Mr protein bands was higher with rennet precipitation as compared with acid precipitation. The rennet coagulation do had a very big Mr protein band at Mr 250kDa. Ammonium sulphate precipitation gave a low yield of whey protein (Lane 5).

Chapter 3, 144

Healthy milk samples

ID SCC (x 103 cells/ml) CMT Score

31 24 Negative

5 6 Negative

4 27 Negative

16 13 Negative

73 27 Negative

110 9 Negative

Min 6

Max 27

Mean 17.67

Median 18.5

Subclinical mastitic milk samples

ID SCC CMT Score

78 4472 2+

75 211 2+

80 482 2+

27 255 2+

97 NA 2+

Min 211

Max 4472

Mean 1355

Median 368.5

Table 3-1: The SCC measurement in all control (n=6) and subclinical mastitic milk samples (n=5) and CMT scores in subclinical mastitic milk samples used in this study for 1D and 2D gel electrophoresis. The range of SCC measurement for healthy milk samples was between 6 and 27 x 103 cells/ml with an average of 18 x 103 cells/ml and median of 19 x 103 cells/ml. The range of SCC measurement for subclinical mastitic milk Chapter 3, 145 samples was between 211 and 4472 x 103 cells/ml with an average of 1355 x 103 cells/ml and median of 369 x 103 cells/ml. The CMT scores for all subclinical mastitic milk samples were 2+ and all control milk samples gave negative results on CMT score. NA means not available.

Chapter 3, 146

Control milk samples

SCC (x 103 ID cells/ml)

31 24

5 6

4 27

16 13

73 27

110 9

43 19

60 80

49 105

110 46

71 43

95 75

Min 6

Max 105

Mean 39.5

Median 27

Chapter 3, 147

Clinical mastitic milk samples

SAA Bacteriological ID Hp (μg/ml) (μg/ml) culture

24 16803.62 20.52 Strep. uberis

9 11277.6 19.92 Strep. uberis

17 559.5 10635.28 No growth

18 6296.3 99.5 No growth

3 2905.1 559.5 E. coli

7 7980.4 115.81 Strep. uberis

78 4419.15 16.57 No growth

47 2815.77 131.34 No growth

A94 1517.24 17.11 No growth

2 4772.38 502.07 E. coli

4 3411.28 42.47 Strep. uberis

6 3628.2 336.46 No growth

Min 559.5 16.57

Max 16803.62 10635.28

Mean 5532.211667 1041.37917

Median 4023.675 107.655

Table 3-2: The SCC measurement in all control (n=12) milk samples and APP concentration in clinically infected milk samples (n=12) used in this study for DiGE. The range of SCC measurement for healthy milk samples was between 6 and 105 x 103 cells/ml with an average of 40 x 103 cells/ml and median of 27 x 103 cells/ml. The range of Hp concentration was between 559.5 and 16803.62 μg/ml with an average of 5532.21μg/ml and median of 4023.68μg/ml. The range of SAA concentrations in clinically infected milk samples was between 16.57 and 10635.28μg/ml with an average of 1041.38μg/ml and median of 107.66μg/ml. From the results of bacteriological culture of Chapter 3, 148 clinical mastitic milk samples, bacteria that had been successfully identified include S. uberis and E. coli although in some milk samples, no growth of bacteria was identified. Chapter 3, 149

Lane 1 Lane 2 Lane 3 Lane 4 Lane 5

250 kDa 150 kDa 100 kDa 75kDa

50 kDa

37kDa

25 kDa

20 kDa

15kDa

10 kDa

Figure 3-1: 1D gel electrophoresis whey samples after various methods of casein precipitation in co3mparison with whole milk sample without any treatment. Mr standard on Lane 1, whole milk on Lane 2, whey sample after rennet coagulation method on Lane 3, whey sample after acid precipitation method on Lane 4 and whey sample after ammonium sulphate fractionation method on Lane 5.

Chapter 3, 150

3.3.3 1D gel electrophoresis on individual subclinical mastitic and healthy milk samples Rennet precipitation was subsequently used to remove major casein protein to allow analysis of lower abundant protein in whey. At the concentration used, rennet did not give a protein band stainable with Coomassie blue (not shown). The separation of proteins from individual subclinical mastitic and healthy milk samples on 1D gel electrophoresis is shown in Figure 3-2a, 3-2b, 3-2c. In comparison of mastitic to healthy milk, no clear difference in terms of presence or absence of protein bands observed (Fig 3-2a). However, despite having no significant difference of protein bands, visualized protein bands in one individual sample of each group (subclinical mastitis in Lane 3; healthy in Lane 10) was digested with trypsin and identified by MALDI-MS. Upon, protein identification searched against Mascot database, only major milk proteins and some serum-derived proteins were successfully identified such as α-lactalbumin, β-lactoglobulin, albumin, lactoferrin and immunoglobulins (Fig 3-2b and Fig 3-2c). These results were similar between subclinical mastitic and healthy milk samples. No protein was successfully identified from the database for high molecular protein bands such as band 1, 2, 11 and 12 or for low Mr protein bands such as 10, 21, 22 (Fig 3-2b). The whole list of successful protein identification by Mascot database is presented on Table 3-3.

3.3.4 2D gel electrophoresis on pooled subclinical mastitic and healthy milk samples The second dimension protein separation on both pooled subclinical mastitic and healthy milk samples after casein removal by rennet precipitation is shown in Figure 3-3a and 3- 3b. The IPG strip used was between pH 3 and pH 10. Upon colloidal Coomassie staining of the gel, there was no clear difference in terms of presence, absence, increase or decrease intensity of protein spots observed between these two pooled milk samples. There was only increased in the intensity of albumin protein spot (Spot 1) seen on 2D gel of pooled subclinical mastitic milk sample in comparison with pooled healthy milk sample. Based on previous study by Hogarth et al. (2004), it was concluded that protein spot 2 and 3 were the major whey proteins; β-lactoglobulin and α-lactalbumin respectively. Thus, no further mass spectrometry analysis was done on this 2D gel.

Chapter 3, 151

Lane 1 Subclinical Mastitis Healthy

250 kDa

150 kDa

100 kDa

75 kDa

50 kDa

37 kDa

25 kDa 20 kDa

15 kDa

10 kDa

Figure 3-2a: 1D gel electrophoresis of 5 individual subclinical mastitic and 6 individual healthy milk samples. Showing protein bands for each of milk samples and the Mr standard was on Lane 1. Chapter 3, 152

Figure 3-2b: 1D gel electrophoresis of a subclinical mastitic and a healthy milk sample. Showing protein bands used for MS analysis. Chapter 3, 153

Figure 3-2c: 1D gel electrophoresis of a subclinical mastitic and a healthy milk sample. Showing all the proteins identified by the MS based on the Mascot database.

Chapter 3, 154

Table 3-3: Proteins identified after in-gel trypsin digestion of protein bands on 1D gel electrophoresis. Mascot peptide database was used to identify the proteins. Mr and pI data were from Mascot database result.

Chapter 3, 155

pH 3 pH 10 Spot 1 250 kDa

Spot 2

Spot 3

10 kDa

Figure 3-3a: 2D gel electrophoresis of pooled subclinical mastitic whey sample. Showing major milk proteins such as albumin on Spot 1, β-lactoglobulin on Spot 2 and α-lactalbumin on Spot 3 (Hogarth et al. 2004). Chapter 3, 156

Spot 4

Spot 5

Figure 3-3b: 2D gel electrophoresis of pooled healthy whey sample. Showing major milk proteins such as β-lactoglobulin on Spot 4 and α-lactalbumin on Spot 5 (Hogarth et al 2004).

Chapter 3, 157

3.3.5 Difference gel electrophoresis (DiGE) on pooled clinical mastitic and healthy milk samples Difference gel electrophoresis was performed on clinical mastitic samples in comparison with healthy samples to increase the likelihood of identifiable difference and after casein removal by rennet precipitation to increase the sensitivity of lower-abundance proteins detection. The scanned gel of the overlay image where both samples labeled with Cy3 and Cy5 fluorescent dyes can be analyzed on the same image as shown in Fig 3-4a. The gel stained with Cy5 dye for the mastitic milk is shown in Fig 3-4b and with Cy3 dye for normal milk is shown in Fig 3-4c. No clear difference was observed in terms of presence, absence, increase or decrease intensity of protein spots between the pooled two milk samples except for increase in the intensity of protein spot of albumin and reduced intensity of the protein spot of whey proteins in mastitic milk sample as compared with healthy milk sample. Otherwise, no other significant protein spot was detected on these gels. Chapter 3, 158

Figure 3-4a: Difference gel electrophoresis on clinical mastitic and healthy milk samples. Showing clinical mastitic milk sample was labeled with Cy5/red fluorescent dye and healthy milk sample was labeled with Cy3/green fluorescent dye. Chapter 3, 159

Albumin

Whey proteins

Figure 3-4b: Difference gel electrophoresis on pooled clinical mastitic milk sample on Cye5 fluorescent dye. Increased in albumin and decreased in whey proteins spot intensities are shown by the arrows as compared with pooled healthy milk sample in Figure 3-4c. Chapter 3, 160

Whey proteins

Figure 3-4c: Difference gel electrophoresis on pooled healthy milk samples on Cye3 fluorescent dye. Increased in whey proteins spot intensity is shown by the arrow as compared with pooled healthy milk sample in Figure 3-4b. Chapter 3, 161

3.4 Discussion The clinical mastitic milk samples chosen in this study had high concentrations of acute phase proteins for both SAA and Hp. The mean values for APP concentrations in clinically infected milk samples were 5532μg/ml and 1041μg/ml for Hp and SAA respectively. The range in values for APP concentrations in clinically infected milk samples were 560 to 16804μg/ml and 16.6 to 10635μg/ml for Hp and SAA respectively. As the distribution of these data was not normally distributed, the median values for each parameter were obtained. The median values for APP concentrations in clinically infected milk samples were 4024μg/ml and 107.7μg/ml for Hp and SAA respectively. Because no APP concentration was obtained for either the healthy or the subclinical mastitic milk samples as these analyses were carried out retrospectively after development of the Hp ELISA assay and the specific samples were no longer available, no significance test was performed. However, the Spearman’s correlation test was performed which revealed a negative and low relationship between SAA and Hp concentrations (r= -0.364). It was surprising and interesting and may reflect different origins of Hp and SAA in milk, notable that recent evidence suggests that Hp in milk may be derived from neutrophils (ie somatic cells) as well as epithelial tissue (Cooray et al. 2007). The SCC measurement for all healthy milk samples was less than 100 000 cells/ml (average was 40 000 cells/ml) which confirmed the healthy status of the cows. The Mann Whitney’s test revealed significant difference (p<0.05) in the SCC measurement of control and subclinical mastitic milk samples with a significance level (p) of p=0.001.

The objective of this study was to characterize the milk proteome associated with bovine mastitis on gel-based proteomic techniques and to characterize and identify low abundance protein in the milk proteome using mass spectrometry. Previous studies as described in chapter one clearly demonstrated that proteomic analyses have successfully characterized and identified several important milk proteins which might be served as protein biomarker of bovine mastitis (Yamada et al. 2002; Smolenski et al. 2007). Furthermore proteomics has also helped to study the physiological functions of those identified proteins in the maintenance of mammary gland in terms of providing nutrients immunity and development to the newborns.

To examine changes in low abundance proteins, rennet coagulation was chosen over acid precipitation or ammonium sulphate fractionation. Based on 1D gel electrophoresis, rennet coagulation successfully removed the majority of milk casein. Furthermore, the intensity of Chapter 3, 162 lower abundance proteins at 50kDa to 250kDa of molecular weight was increased when rennet coagulation was used. However, other major milk proteins such as α-lactalbumin and β-lactoglobulin were still present in the milk samples. Potential interference in protein pattern for the protein of the rennet enzyme was discounted. A rennet precipitation of a similar concentration to that used for casein precipitation was run on 1D SDS PAGE and gave no Coomassie blue stained band.

For a preliminary analysis, 1D gel electrophoresis of individual milk samples was performed on subclinical mastitis in comparison with healthy cases. However, there was no distinctive difference in terms of protein bands between subclinical mastitic and healthy milk samples (Figure 3.2a, b, and c). Furthermore, no lower abundance protein bands were seen in milk samples after casein removal in the healthy or mastitic sample. Unfortunately, the high molecular weight protein bands but with lower intensity were not successfully identified by the MALDI-MS.

Colloidal Coomassie blue staining might not be sensitive enough to detect all lower abundance proteins which may still include an important protein biomarker in bovine mastitis. However, there was no significant difference in terms of protein levels that could differentiate subclinical mastitic from healthy milk sample in this study. Another possible reason was due to the limited usage of 1D gel electrophoresis as it only separates milk proteins based on their molecular weight so that 2DE may be more effective. Low intensity but high molecular weight protein bands (Band 1, 2 ,11 and 12) were not successfully identified due to very low level of proteins exist in those protein bands. Limited number of B. taurus database entries in the database used (Mascot) might contribute to no possible protein identification of these protein bands.

However, for further characterization of milk proteins on 1DE, separated protein bands in a milk sample from each group were cut from the gel and sent for MS analysis. MALDI- TOF successfully identified more of these protein bands though it appeared that the identified proteins were the normal milk proteins that can be found in bovine milk. These include albumin, α-lactalbumin and β-lactoglobulin. This result was in accord with previous proteomic analyses (Hogarth et al. 2004; Smolenski et al. 2007; Boehmer et al. 2008; Boehmer et al. 2010). No casein was identified on the 1DE gel showing that rennet precipitation had successfully removed casein.

Beta-lactoglobulin which comprises about 10% of the total milk protein was identified by MALDI-TOF with Mr of 18554 and pI value of 4.83 for subclinical mastitic milk sample Chapter 3, 163 and slightly higher Mr of 18640 with the pI value of 4.76 for healthy milk sample. Beta- lactoglobulin binds to retinol and thus implicates its function in the transport of vitamin A to newborns (Wong et al. 1996). Alpha-lactalbumin which accounts for about 13% of total whey protein was identified by MALDI-TOF in this study with a Mr of 11541.4 kDa with pI values of 4.61. It acts as a specifier protein which catalyzes the biosynthesis of lactose in addition of galactosyltransferase (Wong et al. 1996). During intramammary infection, there is a possible reduction in the levels of α-lactalbumin and β-lactoglobulin in milk has been reported due to depletion in the synthesis and secretion of these proteins following damage to the mammary epithelial cells by bacterial toxins or by the presence of infection in the mammary gland (McFadden et al. 1988).

It is known that serum-derived proteins such as albumin, immunoglobulins, complement proteins (Ward et al. 2002) and lactoferrin are present in low levels in bovine milk (Lacy- Hulbert et al. 1999; Smolenski et al. 2007). These may have antimicrobial properties (Ward et al. 2002). Albumin is also known to increase in concentration during mastitis as a result of leakage from blood circulation into the mammary gland (Hogarth et al. 2004). It has also been demonstrated that albumin can be synthesized and secreted in the bovine mammary gland as local expression of albumin in the mammary gland increased upon exposure to LPS as well as in mastitic condition (Shamay et al. 2005). However, because no quantification of specific protein levels in milk samples was made, it is not known whether the presence of albumin in subclinical mastitic milk was significantly different from its presence in healthy milk in this study. Other than albumin, immunoglobulins and lactoferrin were also present in both subclinically infected and healthy milk samples. Epithelial cells of the mammary gland are known to produce lactoferrin as well as cytokines (IL-8 and TNF-α) in response to LPS stimulation (McClenahan et al. 2005; Pareek et al. 2005).Study by Chaneton et al. (2008) suggested that different bacteria species may stimulate different mammary gland responses mediated by Lf secretion. On the other hand, immunoglobulins are one of the major proteins in colostrum which confer passive immunity to the newborn (Smolenski et al. 2007) which also serve as defence mechanism against microbial invasion of mammary gland.

As no differences were observed on the gel pattern of 1D gel electrophoresis between subclinical mastitic and healthy milk samples, 2D gel electrophoresis was applied to similar samples. It was hope that lower abundance proteins would be identified because 2D electrophoresis is generally more effective in separating as in bovine milk. Further pools of Chapter 3, 164 milk samples from subclinical and healthy groups were used prior to 2D gel electrophoresis.

Unfortunately, there was again no observable difference in low abundance proteins between the pooled subclinically infected milk and pooled healthy milk samples on the 2D gel. Mass spectrometry analysis was not undertaken as the protein spots in both gels represent the same major milk proteins; albumin, α-lactalbumin and β-lactglobulin and have previously been identified (Hogarth et al. 2004). However, it was interesting to find that albumin was not observed on the 2D gel of healthy samples. This result further supported the findings that albumin was increased in concentration during bovine mastitis as a result of either an increase in leakage from serum into the mammary gland or in albumin expression in the mammary gland. No reduction in the intensity of protein spots of other whey proteins in subclinical sample was shown in this 2D gel.

In this study, no low abundance protein was seen on the 2D gel of subclinical mastitic sample and the possible explanation may derived from the evidence or the potential binding of other proteins to serum albumin. Serum albumin is known to bind as well transport long-chain fatty acids and other small molecules such as bradykinin, interferon and even acute phase proteins (SAA and α-1-acid glycoprotein) (Peters 1977; Gundry et al. 2007). Furthermore, 2D gel electrophoresis was thought to have difficulty to separate low abundance proteins, acidic and basic proteins, proteins with very small or very large in size as well as hydrophobic proteins (Beranova-Giorgianni 2003). This can be explained of why no low abundant protein particularly APP seen on 2D gel of subclinical mastitic sample.

In contrast to studies which have removed casein, it has been suggested that in order to increase the likelihood of the detection of lower abundance proteins, no sample fractionation or depletion of high-abundance proteins should be performed (Boehmer et al. 2008). Although several other studies suggest the use of fractionation or immunoprecipitation (Yamada et al. 2002; Hogarth et al. 2004) to selectively remove casein, β-lactoglobulin, α-lactalbumin or other major milk proteins, the possibility of other lower-abundance proteins also being removed might occur. To investigate subclinical mastitis, more milk samples should also be examined to increase the potential of discovering protein biomarkers important in bovine mastitis. Another way to improve and increase the separation and detection of lower abundance proteins is by using more sensitive staining methods than colloidal Coomassie including silver or sypro-ruby staining (Hogarth et al. 2004). As investigation of milk samples from sub clinical mastitis cases did Chapter 3, 165 not give any significant difference in the levels of milk proteome, the examination of milk samples from clinical mastitis should be considered to find identity possible biomarker after which change in the subclinical mastitis could be determined.

Therefore, milk samples from clinical mastitis cases were collected for comparison with healthy milk samples and analyzed using difference gel electrophoresis (DiGE). DiGE enables the visualization of multiple protein samples in one 2D gel. After labelling the milk samples with the differing fluorescent dyes and running on the same gel, spot patterns were visualized by alternately illuminating the gel with the excitation wavelengths of each dye. Overlay image was then analyzed using standard proteomics software (Van den Bergh et al. 2003).

Unfortunately, although clear distinction were found between clinical mastitic and healthy samples on DiGE gels with lower abundance proteins and increased albumin in the mastitic sample, all of the changes had been observed before (Hogarth et al. 2004). Thus, no further mass spectrometry analysis was done and based on the previous study (Hogarth et al. 2004), the protein spots seen on the Cy dye stained gels were actually the major milk proteins (albumin, β-lactoglobulin and α-lactalbumin).

The protein concentration of the non-casein but still high abundance proteins (albumin or whey proteins) of these milk samples might also contribute to failed detection in other lower abundance proteins. Thus, in order to increase the potential of using difference gel electrophoresis, several important factors must be taken into consideration. The use of more clinical mastitic milk samples may increase the sensitivity to detect lower abundance proteins. Milk samples should be handled more cautiously in terms of transporting the samples from farm to laboratory as well as minimizing the thawing process of milk samples. This would ensure no depletion of important milk proteins should occur.

In conclusion, among milk proteins successfully identified by MALDI-MS on 1D gel electrophoresis were albumin, α-lactalbumin, β-lactoglobulin, lactoferrin and immunoglobulins. However, no clear distinction was observed between the 2D gel electrophoresis of subclinical mastitic and healthy milk samples. Additionally, from DiGE analysis, albumin was found increased in clinical mastitic sample whilst β-lactoglobulin was reduced in mastitic sample when compared with healthy sample. From this study, it was evident that gel-based proteomic techniques as used here were not able to identify novel protein biomarker in milk in bovine mastitis or other lower-abundance proteins involved in inflammation or host defence mechanism. The technical requirement and Chapter 3, 166 resource implication needed for consistent analytical result in a gel-based approach to proteomic analysis of milk are further aspects to consider. Accordingly, non-gel approaches were undertaken to continue the investigation such as CE-MS and LC-MS/MS in the next chapters.

Chapter 4, 167

Chapter IV: Peptidomic Analysis on Bovine Mastitis

4.1 Introduction

Peptidomic analysis involves extraction, separation, detection and identification of peptides in samples. In earlier studies, less manipulation of the samples was employed to allow peptides to be analysed by mass spectrometry (Baggerman et al. 2004). Tissues were directly placed on the MALDI target plate together with the matrix solution (Clynen et al. 2001; Verhaert et al. 2001; Uttenweiler-Joseph et al. 1998; Sweedler et al. 2000) although in the case of ESI, a simple extraction was performed in the spraying solvent (Kollisch et al. 2000). However, due to the presence of multiple biological peptides at different concentrations in cells, tissues or body fluids, it is desirable that peptide extraction and separation are used prior to mass spectrometric analysis. Indeed, the aim of several studies is to enrich the low-abundance peptide composition of samples (Chertov et al. 2005; Baggerman et al. 2002; Skold et al. 2002; Schulz-Knappe et al. 2001).

As the main difference between protein and peptide is based on their molecular weights, the isolation of smaller peptides from higher concentration of larger proteins is essential (Hu et al. 2009). Several extraction steps for analysis of endogenous peptides have been employed to meet this aim. Some methods which have been applied include ultrafiltration, organic solvent precipitation, solid phase extraction, restricted-access material (RAM) - trap column-based extraction and others (Hu et al. 2009).

In ultrafiltration, a membrane with an accurate molecular weight cut-off (MWCO) is used to extract peptides of interest while removing other high molecular weight (HMW >20kDa) peptides and proteins. However, in this method, the ultrafiltration times will increase dramatically if huge amount of sample is applied and also, other LMW (<20kDa) contaminants (e.g. salt) may also be concentrated which may affect the efficiency of the method (Hu et al. 2009).

By using organic solvent precipitation, the electrostatic interactions between molecules are increased and the smaller peptides are dissolved in the high concentration of solvents (Polson et al. 2003). An example of an organic solvent that has been used is acetonitrile (Chertov et al. 2004). The use of solid-phase extraction (SPE) has also been employed as

Chapter 4, 168 part of sample clean-up preparation while functionalized magnetic particles were used as adsorbents to enrich specific peptides in various studies (Villanueva et al. 2004; Fiedler et al. 2007; Baumann et al. 2005). In addition, conventional chromatographic packing materials such as reversed-phase polymers and nanomaterials with different properties have been employed as part of this methodology (Aristoteli et al. 2007; Li et al. 2007; Tian et al. 2009).

All these methods are examples of off-line mode sample preparation which may be laborious, time-consuming and have poor reproducibility (Hu et al. 2009). The use of a special type of chromatographic packing materials called restricted access materials (RAMs) with two functions during chromatographic separation has been proposed. It was suitable for online extraction of smaller peptides and removal of other high abundance peptides (Boos & Grimm 1999). Other techniques including selective electrophoresis and continuous elution electrophoresis have similar characteristics in their ability to concentrate LMW peptides (Camerini et al. 2007; Ly et al. 2008).

The next step of peptidomics involves peptide separation. Baggerman et al. (2004) reviewed several types of peptide separation methodologies that can be employed. One of them is the use of coupling an HPLC to electrospray sources. In this method, the peptides are separated by the use of reversed-phase chromatography (Schoofs et al. 1993) and by applying the voltage difference between the outlet of HPLC and the sample targets, ions can be transferred from solution into the gas phase. This method allowed sensitive peptide analysis and have been applied to various peptidomics studies (Schulz-Knappe et al. 2001).

Recent advances have been developed to detect lower concentrations of peptides from complex protein mixtures. When ultra low flow LC systems coupled to nanospray sources are used, the flow rates employed for chromatographic separation has been as low as 20nl/min with the internal diameters of columns down to 15μm (Shen et al. 2002). However, this method is time consuming and to prevent this, column switching was introduced to separate the sample loading and the actual separation. A whole sample can be analyzed and peptides identified can be selected for further fragmentation in the collision cell of the MS instrument as they passed through the columns (Baggerman et al. 2004). When analysing more complex samples, reducing the flow rate during the separation can be performed to increase the time analysis of the mass spectra and thus, more peptides can be fragmented or fragmented longer in the case of LMW peptides.

Chapter 4, 169

Another type of peptide analysis is multi-dimensional peptide separation. Reversed-phase LC is frequently used as the second separation stage which enables full separation of all the effluent material from the first step. It can be combined with size exclusion reversed- phase (Oosterkamp et al. 1998), reversed phase-reversed phase chromatography (Oosterkamp et al. 1998: Wall et al. 1999) or ion exchange-reversed phase (Oosterkamp et al. 1998; Davis et al. 2001). Multidimensional protein identification technology (MudPIT) using a strong cation resin (SCX) and RP columns has been successfully applied for shotgun proteomic research (Hu et al. 2009). The individual peptides separated from SCX columns were eluted onto the RP columns before subsequently being sprayed into the ESI MS (Baggerman et al. 2004; Hu et al. 2009).

Most peptidomes, unlike tryptic peptides, do not carry C-terminal Lys or Arg residues and SCX column separates peptides based on their charges (Hu et al. 2009). Reversed phase (RP) on the other hand is pH stable, and by using RP-RP separation, peptides may be better separated as shown by Dowell et al. (2008). A total of 56 peptides from known neuropeptides precursors including 17 unidentified peptides, have been successfully separated using this method. A combination of sample ultrafiltration with size-exclusion chromatography prefractionation has been considered as a good method for comprehensive peptidomics analysis (Hu et al. 2007). It allows the investigation of HMW peptides and at the same time, identifying the LMW peptides which improved the sensitivity of LMW method of detection. The advances in the MS analysis in detecting HMW peptides have developed with the use of quadrupole TOF MS with collision-induced dissociation (Möhring et al. 2005), fourier transformion cyclotron resonance MS using electron capture dissociation and linear quadrupole ion-trap MS wit electron transfer dissociation (Hu et al. 2009) which allowed sequencing of large and small peptides.

Another increasingly popular peptidomics analysis is the combination of capillary electrophoresis (CE) with MS. It can perform a separation of peptides from small molecules to large proteins in the same analysis. It has been chosen as a method of analysis due to the low amount of sample used and short analysis time (Baggerman et al. 2004). Pre-concentrating of the sample with hydrophobic media can be employed to increase the concentration of the sample.

Quantitative peptidomics has been increasingly important especially in biomarker discovery when it is important to understand the relative difference of each protein/peptide in different conditions (Hu et al. 2009; Baggerman et al. 2004). Direct profiling with

Chapter 4, 170

MALDI/SELDI is most widely used to quantify peptides but another way is the by using a UV detector between the LC system and the MS to compare the peptide relative abundances (Hu et al. 2009; Baggerman et al. 2004).

Differential peptide display was introduced by Tammen et al. (2003) in which the sample was separated by RP-HPLC and each fraction eluted was analysed by MALDI-TOF MS. The mass spectra were transformed into 1D gel-like pattern and the patterns were combined to generate a virtual 2D peptide map. To identify the peptides, individual HPLC fractions were subjected to nESI-qTOF-MS/MS or MALDI-TOF/TOF-MS sequencing.

The SELDI-TOF MS technique is another method using the protein chip array technology with a modified surface target (Huchens & Yip 1993). The target surface can be modified with chemical functionality such as RP, ion exchange chromatography (IEX), immobilized metal affinity chromatography, antibodies, proteins and DNA. This binding activity was considered as a pre-fractionation step to allow easier analysis. Different expressed peptides can be identified from different groups of sample using advanced bioinformatic tools (Issaq et al. 2003; 2002).

However, these methods described above (MALDI/SELDI profiling) which were thought to provide semi-quantitative analyses of peptides as ion patterns, could be affected by ionization efficiency, ion suppression effects and sample preparation process (Hu et al. 2009). Furthermore, ion intensities in the mass spectrum did not correlate to the relative amounts of the corresponding compounds (Baggerman et al. 2004). To overcome this problem, mass labelling was employed in which a known concentration of a „mass labelled‟ analogue of the peptide of interest was used as internal reference. A perfect example of mass labelling was by incorporating stable isotopes in which two different samples had the same ionization efficiency but with a slight difference in masses (Hu et al. 2009; Baggerman et al. 2004).

An iTRAQ approach with amine reactive isobaric tagging and tandem MS to quantify the peptides has been employed in both proteomics and peptidomics analyses. Peptides with the same sequence appeared to be identical in masses but produced as a single ion signal in MS (Hu et al. 2009). It may be used in complementary studies with MALDI profiling for both clinical diagnosis and biomarker discovery as performed by Hu et al. (2009).

Another isotope labelling method used for cysteine containing peptides/proteins is isotope- coded affinity tags (ICAT). The side chains of cysteinyl residues of protein are derivatized

Chapter 4, 171 with an isotopically light form of the ICAT reagent. By comparing their relative signal intensities, the relative quantification of samples in different conditions can be determined (Gygi et al. 1999).

In a previous discussion about peptide analysis in bovine mastitis, most of the studies identified proteolytic activities yielding peptides in milk (Wedholm et al. 2008; Larsen et al. 2010). Many diseases are associated with proteolytic processes (Hu et al. 2009). Proteases reflect the response of the host towards the different physiological conditions. Different disease states will lead to different degradation pattern of proteins and these degradation products may be of useful as disease biomarkers (e.g. cancer) as described by Villanueva et al. (2006). The activity of proteases in the biological samples can be quantified by the use of activity-based probes (ABPs) coupling to MS. ABP work by targeting specific proteases based on their enzymatic mechanisms such as serine proteases, cysteine proteases, threonine proteases, aspartyl proteases and metalloproteases (Kidd et al. 2001; Greenbaum et al. 2000, Greenbaum et al. 2002, Kessler et al. 2001, Chun et al. 2004; Saghatelian et al. 2004). Affinity chromatography can also be used to characterize proteases as performed by Tian et al. (2006).

Peptidomic analysis allows full visualization and characterization of the whole peptidome in given cells, tissues or bodily fluid and even their post-transitional modifications peptides. With the use of nanoscale LC coupled with MS, one can identify peptides at the acid amino sequence level however, the use of tandem MS can only provide partial amino acid sequences of peptides. This can be resolved by using certain bioinformatic programmes to identify peptides from protein database or analogue gene from the genomic database (Choudhary et al. 2001). Peptidomics analysis can be applied in complementary with other protein characterization methodologies to discover novel disease biomarkers as well as to identify newly discovered peptides for therapeutic purposes.

4.1.1 Aims of study

Overall, the objective of this chapter is to apply a peptidomic approach to study the presence of polypeptides that may be used to discriminate normal milk samples from milk samples from quarters infected with bacteria causing bovine mastitis. It was hoped that by using this sensitive and high-throughput methodology, significant results can be achieved and lead to the discovery of important and significant mastitis biomarkers. Milk samples

Chapter 4, 172 from healthy cows were compared to samples from cows with clinical mastitis caused by either S. aureus or E. coli. Initially samples were analyzed by CE-MS to detect positive biomarkers and then by LC-MS/MS to sequence putative biomarkers.

4.2 Materials and Methods

4.2.1 Milk Samples and Reagents Milk samples from clinical mastitic cases were obtained from Vale Veterinary Laboratory, Devon, UK. Bacteriological culture was performed in the Bacteriology Laboratory of Vale Veterinary Laboratory to determine the bacterial cause of bovine mastitis. Milk samples from mastitis caused by either E. coli (n=10) or S. aureus (n=6) infection were selected for further analysis. Control milk samples (n=10) from healthy cows from the Cochno Farm of School of Veterinary Medicine, University of Glasgow were also used in this study as comparison to those clinical mastitic cases. The healthy milk samples were confirmed by somatic cell counts (<100 000 cells/ml). Prior to peptidomics analysis, SAA and Hp assays (Chapter 2) were performed determine the concentration of both acute phase proteins in each milk sample. Samples were stored at -20°C prior to analysis.

Buffers and reagents for both sample preparation and CE-MS analysis are presented as in Appendix.

4.2.2 Milk sample preparation for CE-MS using PD-10 Desalting Column

4.2.2.1 Preparation of fresh ammonium (NH4OH) and urea buffer

An aliquot of 25% NH4OH (400μl) was added into 1 litre of deionised water to make up

NH4OH buffer. The solution was mixed by inversion and pH value was checked (pH 10.5- 11.5). Urea buffer was prepared by mixing 5ml of 8M urea, 0.4ml of 5M NaCl, 10μl of

25% (v/v) NH4OH solution and 100μl of 2% (w/v) SDS solution.

4.2.2.2 Sample Preparation Milk samples were defrosted at room temperature prior to centrifugation at 14 000 x g for 10 minutes at 4ºC to remove the fat. Two aliquots of 1ml milk samples were prepared. An aliquot of 700μl of centrifuged milk was added to 700μl of urea buffer, placed in a centrisart tube (Sartorius, Goettingen, Germany) and mixed well by pipetting up and down.

Chapter 4, 173

The filter was put on top of the centrisart tube and centrifuged at 3400 rpm for 80 minutes at 4ºC. During centrifugation, a PD-10 column (GE Healthcare, Sweden) was equilibrated with 25ml of NH4 solution. After centrifugation, 1.1ml of centrisart filtrate was added into the columns until the column was dry. Next, 1.9ml of NH4 buffer was added into columns until the column was dry. The column was then placed on 15ml collecting tubes and the sample was eluted with 2ml of NH4 buffer. The 2ml of flow-through was distributed between three Eppendorf tubes (2 tubes containing 900 µl to be used for CE and LC analysis and 1 tube for protein determination containing 150 µl). Tubes to be used for CE and LC analysis were freeze-dried until further use.

4.2.3 Protein concentration Protein concentration in milk extracted samples was checked using the BCA (bicinchoninic acid) assay Uptima from Interchim (Montluçon, France), About 25μl of each standard and milk extracted samples were pipetted into microplate wells in duplicates. The BCA assay reagent was then added at 200μl and further mixed. The sample mixtures were incubated at 37°C for 30 minutes or 2 hours at room temperature. Prior to reading the results at the optical absorbance of 562 nm, the microplate was cooled at room temperature. The standard curve was plotted and the protein concentration in the sample was interpolated. Freeze-dried aliquots have to be resuspended in appropriate volume of Milli Q water to reach a concentration of 2μg/μl before running in the CE.

4.2.4 CE-MS Analysis

4.2.4.1 Preparation of buffers for CE-MS analysis All buffers were freshly prepared every week. 1M sodium hydroxide (NaOH) was prepared in deionised water. Ammonium hydroxide buffer was prepared by adding 3.76 ml of ammonium solution (25%) to deionised water and the volume adjusted to 50 ml with deionised water. Running buffer was prepared by adding 10 ml of acetonitrile and 472 μl of formic acid to a final volume of 50 ml with deionised water. The sheath flow liquid was prepared by mixing 15 ml of 2-propanol and 200 μl of formic acid with the volume adjusted to 50 ml with deionised water. The standard protein/peptide solution (0.5 pmol/μl) for the calibration of CE-MS analysis contains Lysozyme (14,303 Da, L4919, Sigma-Aldrich, Dorset, UK), ribonuclease (13,681 Da, R5500, Sigma-Aldrich, Dorset, UK), aprotinin (6,513 Da, A1153, Sigma-Aldrich, Dorset, UK) and 4 synthetic peptdies : ELMTGELPYSHINNRDQIIFMVGR (2,832 Da), TGSLPYSHIGSRDQIIFMVGR (2,333 Da), GIVLYELMTGELPYSHIN (2,048 Da), REVQSKIGYGRQIIS (1,733 Da).

Chapter 4, 174

4.2.4.2 CE-MS Analysis for peptide fingerprint The capillary was carefully installed into the CE system (Beckman Coulter, Fullerton, USA). The capillary was first conditioned with NaOH (pressure 50 psi) for 15 minutes.

The capillary was then washed with NH4OH solution (50 psi) for 15 minutes and 20 minutes with running buffer (50 psi). The capillary was not connected to the MS until the conditioning procedure has been completed The Beckman CE was set to run in “reverse mode” when connected to the MS system using the external detector adapter. Running buffer (pressure applied 50 psi) was used to rinse the capillary prior to each injection. Freeze dried samples were re-suspended to achieve a final concentration 2 µg/µl with HPLC water and centrifuged at 14,000 x g for 10 min at 4 ºC shortly before analysis. The samples were then injected at a pressure of 2.0 psi for 99 s resulting in a loading volume of 290 nl of sample volume. Separation was done with +25kV at the injection site for 30 minutes with a capillary set temperature of 35°C. In addition to the +25kV, pressure is applied as followed ( 0.1 psi for 1 min, 0.2 psi for 1 min, 0.3 psi for 1 min, 0.4 psi for 1 min and 0.5 psi for the following 30 min). The sheath liquid was applied coaxially at a running speed of 0.02 ml/h. The capillary was rinsed with deionised water (50 psi) for 1 minute after each run followed by a washing step with NH4OH solution (50 psi) for 3 minutes and lastly by a flushing step with deionised water (50 psi for 3 min). ESI sprayer was grounded to achieve electric potential zero and the electrospray interface potential was set between -4 000 and -5 000 V. The MS spectra were accumulated every 3 s, over a mass-to-charge range from 400 to 3000 m/z for about 60 minutes depending on the analysis requirements. The mass calibration of the MS was performed on a weekly basis using the standard protein/peptide solution (0.5pmol/μl) for CE-MS analysis. CE-MS analysis was performed under supervision of Dr. Bill Mullen and Dr. Amaya Albalat of Institute of Cardiovascular and Medical Sciences, University of Glasgow, UK.

4.2.5 CE-MS Data processing Mass spectral ion peaks were deconvoluted into single masses using MosaiquesVisu software (www.proteomiques.com). Only signals observed in a minimum of 2 consecutive spectra with a signal-to-noise ratio of at least 4 were considered. Signals with a calculated charge of +1 were automatically excluded to minimize interference with matrix compounds or drugs. MosaiquesVisu uses a probabilistic clustering algorithm and both isotopic distribution and conjugated masses for charge-state determination of the entities. CE migration time and ion signal intensity were normalized because migration time and ion-counts showed significant variability between samples mainly due to different salt and

Chapter 4, 175

peptide concentrations. Reference signals of over 380 milk entities or „milk housekeeping polypeptides‟ were used for CE-time calibration by local regression. The same peptides were used for ion signal intensity normalization by a applying a global linear regression. The resulting peak list contained the molecular mass (Da) and normalized CE migration time (min) for each feature. Normalized signal intensity can be used as a measure of relative abundance. Data sets were accepted only if the following quality control criteria were met: A minimum of 950 chromatographic features (mean number of features minus one standard deviation) must be detected with a minimal MS resolution of 8000 (required resolution to resolve ion signals with z=6) in a minimal migration time interval (the time window, in which separated signals can be detected) of 10 min. After calibration, the mean deviation of migration time (compared to reference standards) must be below 0.35 min. Control and disease-specific polypeptide patterns were generated using support vector machine (SVM)-based MosaCluster software.

4.2.6 Statistical analysis Only polypeptides that were found in more than 70% of the samples in at least one of the two groups (controls or cases) were considered for biomarker definition. The pre-defined set of polypeptides should be further validated by randomly excluding 30% of available samples. This was repeated (bootstrapping procedure) up to 10 times. For multiple testing corrections, P values were corrected using the stringent maxT test or using false discovery rate adjustments of Benjamini-Hochberg (BH). Two models were generated for the classification of the samples depending on their CE-MS profile. The first model was used to classify the milk sample as control (healthy) or bacteria (infected) and the second one to discriminate the infectious microorganism (E .coli or S. aureus).

4.2.7 LC-MS/MS Analysis for Peptide Sequencing

4.2.7.1 LC-MS/MS analysis LC-MS analysis was performed using an Ultimate3000 nano-LC pump (Dionex, Sunnyvale, CA, USA) coupled to a LTQ-Velos (Thermo Corp.). Five μL from each sample were loaded onto a Dionex 100μm x 2 cm 5 μm C18 nano trap column at a flowrate of 5 μl/min by a Ultimate 3000 RS autosampler (Dionex, Camberley UK) and separated on a C18 reverse phase column (75 µm ID × 150 mm, 100 Å, Acclaim PepMap RSLC C18, Dionex). Peptides were eluted using a gradient of 1-5-40% B (80% ACN, 0.2% formic acid) in 0-5-60 min at a flow rate of 300 nL / min. The eluant from the column was

Chapter 4, 176 directed to a Proxeon nano spray ESI source (Thermo Fisher, Hemel, UK) operating in positive ion mode then into an Orbitrap Velos FTMS. The ionisation voltage was 2.5 kV and the capillary temperature was 200 ºC. The mass spectrometer was operated in MS/MS mode scanning from 2=380 to 2000 amu. The top 20 multi charged ions were selected from each full scan for MS/MS analysis, the fragmentation method was HCD at 35% collision energy. The ions selected for MS2 using data dependent method with a repeat count of 1 and repeat and exclusion time of 15 s. Precursor ions with a charge state of 1 were rejected. The resolution of ions in MS1 was 60,000 and 7,500 for HCD MS2. LC- MS/MS analysis was performed under supervision of Dr. Bill Mullen and Dr. Amaya Albalat of Institute of Cardiovascular and Medical Sciences, University of Glasgow, UK.

4.2.7.2 LC-MS/MS – Data analysis Raw spectral data from LC-MS/MS analysis of the samples were uploaded to Thermo Proteome Discoverer 1.2. Peak picking was performed under default settings for Fourier transform mass analyzers (FTMS) analysis i.e. only >+2 peptides were considered with signal to noise ratio higher than 1.5 and belonging to precursor peptides between 700 – 8000 Da. Peptide and protein identification was performed with SEQUEST algorithm. An in house compiled database containing proteins from the latest version UniProt SwissProt database was compiled to include only Bos taurus, E. coli and S. aureus entries. This concatenated database was selected to facilitate peptide identification and reduce the probability for false positive identifications upon making searches against 3 differently sized FASTA databases. No enzyme cleavage was selected and oxidation of methionine and proline were chosen as variable modifications. Precursor tolerance was set at 20 ppm and 0.1 Da for MS/MS fragment ions. Resulting peptides and protein hits were further screened by excluding peptides with an error tolerance higher than 10 ppm and by accepting only those hits listed as high confidence by Proteome Discoverer software.

4.3 Results

4.3.1 SCC measurement and acute phase proteins concentration In order to confirm the disease status of milk samples used in this study, SCC measurement and acute phase proteins (SAA and Hp) concentrations were determined. Table 4-1 showed the SCC measurement and APP concentrations in all control milk samples used in this study. In control milk samples, the range of SAA concentration was between 0.079 and

Chapter 4, 177

2.96μg/ml. The average of SAA concentration was 0.71μg/ml and the median of SAA concentration was 0.48μg/ml in control milk samples. Meanwhile, the range of Hp concentration was between 1.95 and 32μg/ml. The average of Hp concentration was 12.8μg/ml and the median of Hp concentration was 12.64μg/ml. The range of SCC measurement in control milk samples was between 10 and 96 x 103 cells/ml. The average was 45 x 103 cells/ml and the median was 46 x 103 cells/ml. Table 4-2 (a and b) showed the bacteriological culture results and the concentration of both SAA and Hp for mastitic milk samples caused by E. coli and S. aureus. For E. coli infected group, the range of SAA concentration was between 4.18 and 1500μg/ml. The average of SAA concentration was 626.6μg/ml and the median of SAA concentration was 90.79μg/ml in E. coli infected group milk samples. Meanwhile, the range of Hp concentration was between 16.03 and 354.26μg/ml. The average of Hp concentration was 112.5μg/ml and the median of Hp concentration was 98.85μg/ml. For S. aureus infected group, the range of SAA concentration was between 16 and 1500μg/ml. The average of SAA concentration was 526.1μg/ml and the median of SAA concentration was 51.54μg/ml in E. coli infected group milk samples. Meanwhile, the range of Hp concentration was between 22.15 and 118.83μg/ml. The average of Hp concentration was 71.44μg/ml and the median of Hp concentration was 78.49μg/ml. The Statistical analysis (Mann Whitney) showed that both SAA and Hp were significantly higher (p=0.00; p<0.05) in the infected group (E. coli and S. aureus) but not significantly different between two bacterial groups (p=0.958; p>0.05 for SAA and p=0.428; p>0.05 for Hp).

Chapter 4, 178

Control milk samples

Sample Ids SAA Conc.(ng/ml) Hp conc. (ng/ml) SCC (x103) 102 0.36 7.78 13 66 0.57 16.38 14 3 0.55 27.98 48 62 0.079 1.95 31 72 1.21 7.24 96 109 0.65 13.21 10 80 0.28 24.61 54 100 0.48 15.36 38 16 0.9 2.86 58 58 0.34 1.95 56 111 0.13 21.94 28 48 0.39 31.95 85 4 2.96 4.17 63 2 1.43 1.95 32 29 0.29 12.64 46 Min 0.08 1.95 10 Max 2.96 31.95 96 Mean 0.71 12.8 45 Median 0.48 12.64 46

Table 4-1: The acute phase proteins (SAA and Hp) concentrations and SCC measurement in all control milk samples used in this study. The range of SAA concentrations was between 0.08 and 2.96 μg/ml with an average of 0.71μg/ml and a median of 0.48μg/ml. The range of Hp concentrations was between 1.95 and 32μg/ml with an average of 12.8μg/ml and a median of 12.64μg/ml. The range of SCC measurement was between 10 and 96 x 103 cells/ml with an average of 45 x 103 cells/ml and a median of 46 x 103 cells/ml.

Chapter 4, 179

Mastitic milk samples SAA Conc. Hp conc. Bacteriological Sample Ids (ng/ml) (ng/ml) results 475 72.13 89.68 E. coli 76 1500 115.18 E. coli ? 1500 78.51 E. coli 814 1500 141.26 E. coli 692 9.48 60.27 E. coli 697 1500 144.89 E. coli 519 109.44 108.02 E. coli 453 4.18 16.63 E. coli 496 24.31 16.03 E. coli 1222 46.02 354.26 E. coli Min 4.18 16.03 Max 1500 354.26 Mean 626.56 112.47 Median 90.79 98.85

Table 4-2a: The acute phase proteins (SAA and Hp) concentrations and bacterial causing mastitis in infected milk samples caused by E. coli used in this study. The range of SAA concentrations was between 4.18 and 1500μg/ml with an average 626.6μg/ml and a median of 90.79μg/ml. The range of Hp concentrations was between 16.03 and 354.26μg/ml with an average of 112.5μg/ml and a median of 98.85μg/ml.

Chapter 4, 180

Mastitic milk samples SAA Conc. Hp conc. Bacteriological Sample Ids (ng/ml) (ng/ml) results 438 48.12 70.77 S. aureus 14 37.21 42.69 S. aureus 980 1500 111.83 S. aureus 708 54.96 95 S. aureus 346 1500 86.2 S. aureus 715 15.99 22.15 S. aureus Min 15.99 22.15 Max 1500 111.83 Mean 526.05 71.44 Median 51.54 78.49

Table 4-2b: The acute phase proteins (SAA and Hp) concentrations and bacterial causing mastitis in infected milk samples caused by S. aureus used in this study. The range of SAA concentrations was between 15.99 and 1500μg/ml with an average of 526.1μg/ml and a median of 51.54μg/ml. The range of Hp concentrations was between 22.15 and 111.8μg/ml with an average of 71.44μg/ml and a median of 78.49μg/ml.

Chapter 4, 181

4.3.2 Polypeptide Fingerprinting for control and infected groups Initially in this work the possibility of differences existing in the milk proteome of healthy cows (n=10) when compared to mastitic cows, regardless of their bacterial cause (n=16) was examined. The polypeptide fingerprints of these groups were very different as shown in Figure 4-1. The y-axis represents the molecular mass (kDa) on a logarithmic scale and the x-axis represents the CE migration time (min) whilst the z-axis represents the mean signal intensity. From the figure, higher molecular mass of peptides appeared to be in higher signal intensity eluting between 25 and 40 minutes in control milk samples compared with infected milk samples. Lower molecular mass of peptides were found to be present in infected milk samples eluting between 50 and 60 minutes although in low signal intensity as compared in control milk samples. Peptides with molecular mass between 1.5 and 4 kDa at 30 to 45 minutes were found to be in higher abundance with higher signal intensity in infected milk samples compared with control milk samples. The next step was to investigate the differences in milk proteome of infected samples caused by two different bacterial pathogens (E. coli and S. aureus). The polypeptide profiles of these 2 different groups were almost similar in pattern as shown in Figure 4-2 except in their signal intensities. Peptides with molecular mass between 2.5 and 3.0 kDa at 25 to 30 minutes were found to be in higher signal intensity in E. coli infected milk samples as compared with S. aureus infected milk samples. Peptides with molecular mass between 2.0 and 4 kDa at 35 and 40 minutes were also found to be in higher signal intensity in E. coli infected milk samples as compared with S. aureus infected milk samples.

Chapter 4, 182

Control Infected (all)

Figure 4-1: The polypeptide fingerprints of both non-infected and infected groups. From these fingerprints, they clearly showed a significant differences in the pattern of polypeptide peaks between non-infected and infected groups. X-axis referred to the CE migration time in minutes whilst y-axis referred to the molecular mass (kDa) on a logarithmic scale and z-axis referred to mean signal intensity.

E. coli S. aureus

Figure 4-2: The polypeptide fingerprints of both E. coli and S. aureus groups. From these fingerprints, they clearly showed differences in the pattern of polypeptide peaks between E. coli and S. aureus groups.

Chapter 4, 183

4.3.3 Model Generation of the polypeptide fingerprints for control and infected groups Based on the polypeptide fingerprints described above, multiple testing of statistical analysis was performed to identify peaks which discriminate between control and infected groups. In the first model, discriminatory peptides were selected that were able to differentiate between the control and infected groups with the area under curve (AUC) set at more than 0.99, Benjamini-Hochberg (BH) was set at less than 0.05 and Tmax less than 0.05. With these very stringent conditions a group of 31 peaks were identified that can be used to discriminate between control and infected groups. From the statistical analysis, this model possessed 100% value for its sensitivity, specificity and accuracy. The peptides that from this model could be used to discriminate between non-infected and infected group is presented in Figure 4-3.

Based on the polypeptide fingerprints described above, multiple testing of statistical analysis was performed to discriminate peaks between two infected groups caused by E. coli and S. aureus. There were three models generated with different values of sensitivity, specificity and accuracy. For Model 1 (Figure 4-4), the area under curve (AUC) was set at more than 0.99, Benjamini-Hochberg (BH) was set at less than 0.05 and Tmax at less than 0.05 and 6 peaks were identified that can be used to discriminate between E. coli and S. aureus groups. The sensitivity value for this model was 100% whilst the specificity value was 91.67% and the accuracy value was 96.3%. For Model 2 (Figure 4-5), with the area under curve (AUC) set at more than 0.8, Benjamini-Hochberg (BH) was set at less than 0.05 and Tmax less than 0.05, 3 peaks were identified that can be used to discriminate between E. coli and S. aureus groups. The sensitivity value for this model was 100% whilst the specificity value was 83.3% and the accuracy value was 92.52%. For the Model 3 (Figure 4-6), with the area under curve (AUC) set at more than 0.9, Benjamini-Hochberg

(BH) was set at less than 0.1 and Tmax less than 0.05, 13 peaks were identified that can be used to discriminate between E. coli and S. aureus groups. The sensitivity value for this model was 100% whilst the specificity value was 83.3% and the accuracy value was 92.52%. These three models with different number of polypeptides were used to discriminate between E. coli and S. aureus infected groups. All polypeptides generated from all three models were tabulated in Table 4-3. It should be noted that a total of 13 polypeptides appeared in at least one of the models (Table 4-4).

Chapter 4, 184

Figure 4-3: The model of polypeptide fingerprint after multiple testing of non- infected and infected groups. This model has successfully identified 31 polypeptide peaks that can be used to discriminate between non-infected and infected groups.

E. coli S. aureus

Figure 4-4: The first model of polypeptide fingerprint after multiple testing of E. coli and S. aureus groups. This model has successfully identified 6 polypeptide peaks that can be used to discriminate between E. coli and S. aureus groups.

Chapter 4, 185

E. coli S. aureus

Figure 4-5: The second model of polypeptide fingerprint after multiple testing of E. coli and S. aureus groups. This model has successfully identified 3 polypeptide peaks that can be used to discriminate between E. coli and S. aureus groups.

E. coli S. aureus E. coli

Figure 4-6: The third model of polypeptide fingerprint after multiple testing of E. coli and S. aureus groups. This model has successfully identified 13 polypeptide peaks that can be used to discriminate between E. coli and S. aureus groups.

Chapter 4, 186

Model 1 Model 2 Model 3 Peptide ID Peptide ID Peptide ID 8093 9662 9662 8098 8093 8093 9092 6503 6503 9662 8098 13428 9092 14889 13428 14889 9960 13328 4865 7557 13101 3255

Table 4-3: Polypeptides that were identified from 3 models generated to discriminate E. coli from S. aureus infected groups. These peptides appeared in at least one of the models

Chapter 4, 187

Peptide Models from where the peptide was ID found 9662 1,2,3 8093 1,2,3 6503 1,2 8098 2,3 9092 2,3 13428 2,3 14889 2,3 9960 3 13328 3 4865 3 7557 3 13101 3 3255 3

Table 4-4: A total of 13 polypeptides generated from 3 models that discriminate E. coli from S. aureus infected groups. Models from where the peptide was found were also listed.

Chapter 4, 188

4.3.4 CE-MS Analysis A total of 31 discriminatory polypeptides were identified from CE-MS analysis and were significant on the basis of comparison between control and infected groups (S. aureus and E. coli) and are presented in Table 4-5. Meanwhile, a total of 14 discriminatory polypeptides were identified from CE-MS analysis and were significant on the basis of comparison between two infected groups (S. aureus and E. coli) and are presented in Table 4-6. A total of 45 discriminatory polypeptides were presented in Table 4-7 based on the sample type comparison (e.g: control vs infected; E. coli vs. S. aureus). These polypeptides were found in more than 70% of the samples in at least one of the two groups (controls or cases) in order to be considered as disease biomarkers.

Chapter 4, 189

Peptide ID Peptide Mass CE Migration time 386 914.5 32.3 895 1013.5 32.1 963 1028.6 28.9 2367 1275.7 32.3 2664 1335.7 37.8 2739 1350.7 25.1 2740 1350.7 27.6 2757 1352.8 21.1 2759 1353.7 29 3422 1481.8 39.2 3728 1554.8 39.2 3913 1597.8 28 4226 1667.9 40 5064 1893 30.8 5217 1933.1 30.2 5263 1947.1 30.8 6017 2153.2 24.4 6178 2197.3 26.7 6465 2277.3 27.6 6891 2390.2 33.1 7095 2440.4 24.8 7323 2490.4 29.7 7333 2493.4 24.8 11937 3720 40.5 14110 4498.4 36.9 14701 4780.6 26.4 16632 6376 41.9 16728 6517.8 29.7 17239 7687.5 40 19020 15873.3 26.6 19046 16123.3 27.8

Table 4-5: The discriminatory polypeptides identified from CE-MS analysis that were significant in the basis of comparison between control and infected groups. A total of 31 polypeptides are presented with their mass values and CE migration times.

Chapter 4, 190

Peptide ID Peptide Mass CE Migration time 3255 1451.8 21.4 4865 1839 27.7 6503 2285.2 33.1 7557 2552.3 27.7 8093 2676.5 34.4 8098 2677.4 34.2 9092 2896.5 41.4 9094 2896.6 59.4 9662 3028.5 39.5 9960 3113.6 27.7 13101 4120.2 37.5 13328 4205.2 28.3 13428 4237.2 37.5 14889 4882.5 38.4

Table 4-6: The discriminatory polypeptides identified from CE-MS analysis that were significant in the basis of comparison between two infected groups (E. coli and S. aureus). A total of 14 polypeptides are presented with their mass values and CE migration times.

Chapter 4, 191

Peptide ID Peptide Mass CE Migration Time Type of comparison 386 914.486 32.3 Control vs Infected 895 1013.55 32.1 Control vs Infected 963 1028.58 28.9 Control vs Infected 2367 1275.67 32.2 Control vs Infected 2664 1335.7 37.8 Control vs Infected 2739 1350.71 25.1 Control vs Infected 2740 1350.71 27.6 Control vs Infected 2757 1352.76 21.1 Control vs Infected 2759 1353.69 29.0 Control vs Infected 3422 1481.8 39.2 Control vs Infected 3728 1554.82 39.2 Control vs Infected 3913 1597.81 28.0 Control vs Infected 4226 1667.9 40.0 Control vs Infected 5064 1893.01 30.8 Control vs Infected 5217 1933.11 30.2 Control vs Infected 5263 1947.07 30.8 Control vs Infected 6017 2153.17 24.4 Control vs Infected 6178 2197.26 26.7 Control vs Infected 6465 2277.28 27.6 Control vs Infected 6891 2390.21 33.1 Control vs Infected 7095 2440.39 24.8 Control vs Infected 7323 2490.36 29.7 Control vs Infected 7333 2493.36 24.8 Control vs Infected 11937 3720.03 40.5 Control vs Infected 14110 4498.43 36.9 Control vs Infected 14701 4780.55 26.4 Control vs Infected 16632 6376.03 41.9 Control vs Infected 16728 6517.75 29.7 Control vs Infected 17239 7687.52 40.0 Control vs Infected 19020 15873.3 26.6 Control vs Infected 19046 16123.3 27.8 Control vs Infected 3255 1451.843 21.4 E. coli vs S. aureus 4865 1838.965 27.7 E. coli vs S. aureus 6503 2285.208 33.1 E. coli vs S. aureus 7557 2552.311 27.7 E. coli vs S. aureus 8093 2676.482 34.4 E. coli vs S. aureus 8098 2677.406 34.2 E. coli vs S. aureus 9092 2896.512 41.4 E. coli vs S. aureus 9094 2896.564 59.4 E. coli vs S. aureus 9662 3028.54 39.5 E. coli vs S. aureus 9960 3113.574 27.7 E. coli vs S. aureus 13101 4120.18 37.5 E. coli vs S. aureus 13328 4205.192 28.3 E. coli vs S. aureus

Chapter 4, 192

13428 4237.227 37.5 E. coli vs S. aureus 14889 4882.521 38.4 E. coli vs S. aureus

Chapter 4, 193

4.3.5 Peptide sequencing In order to identify the discriminatory peptides of the models by their sequences, samples were analysed using LC-MS/MS. Twelve of the 45 polypeptides identified with high confidence are listed in Table 4-8. Based on their masses from the CE-MS analysis and the number of basic amino acids in the sequence (H, R, K or histidine, arginine and lysine), these parameters (peptide mass and number of amino acids) were used to link the peptides obtained in both systems (CE-MS and LC-MS). Figure 4-7 showed a scatterplot of polypeptides in all milk samples from CE-MS analysis based on the number of basic amino acids found in the sequence. The y-axis represents the molecular mass (kDa) of milk polypeptides whilst the x-axis represents the CE migration time (min). Based on the molecular mass and the CE migration time of certain polypeptides, the number of amino acids found in the sequence can be determined from Figure 4-7. The number of basic amino acids can be used in the process of determining the peptide sequence for each polypeptide identified from CE-MS analysis. From Thermo Proteome Discoverer software and SEQUEST algorithm, the potential amino acids sequences were obtained according to these polypeptide masses and number of basic amino acid present in the sequence as presented in Table 4-8. In this table, other important parameters are presented such as the amino acid sequence, protein accession, probability/XCorr/scores, charge, mass-charge ratios (m/z), the protonated mass of the peptide (MH +), DM (ppm), LC migration time (RT in minutes) and number of ions matched. As the methionine and proline oxidations were set as variable modifications you, the lower cases were presented (e.g. p and m) instead of their upper cases (e.g. P and M) in the amino acid sequence. The protein accessions obtained from the software programme are the coded name for the protein from which the peptide belongs and this information was used to search for protein identification using protein database (e.g. UniProt or NCBI; see section 4.3.6 below). The probability, XCorr and score values were the statistical values provided by the Proteome Discoverer and SEQUEST algorithm that depict the validity of the identification. Meanwhile, the charge column in the table referred to the charge state in which the peptide was detected by the mass spectrometer and DM (ppm) means the error difference between the theoretical mass value of the peptide and the experimental mass value from the mass spectrometer. Theory mass was the mass deviation of the peptide which was calculated from the sequence whilst ions matched referred to the produced peptide fragments from air collision in the MS and were recorded in a new mass spectrum to produce the amino acid sequence.

Chapter 4, 194

It should be noted that not all the identified polypeptides from CE-MS analysis were successfully sequenced. Furthermore, even with the same peptide sequence, the statistical values (probability, XCorr, score values) and other parameters appeared to be different as the respective peptide originated from different type of samples (control, E. coli or S. aureus) as happened in Peptide ID 2367, 3728, 3913, 4226, 4865, 6017, 6178 and 6465.

4.3.6 Protein identification As described above, protein identity was determined from a protein database based on their amino acid sequences. The identified and named proteins are presented as in Table 4-9. The UniProt Knowledgebase (UniProtKB) database was used to determine the name of protein by using the protein accession information of each amino acid sequence. These identified proteins will be discussed further in the discussion section.

Chapter 4, 195

ID Protein Theory ΔM Ions Origins of Pb XC Sc C m/z [Da] MH+ [Da] RT [min] Peptide Sequence Accessions Mass [ppm] Matched polypeptides 2367 PFPEVFGKEKV P02662 2.92 3.00 1.00 2 638.85034 1276.69341 1275.68613 -0.21 64.93 11/20 Found in S. aureus sample PFPEVFGKEKV P02662 6.16 3.97 1.00 3 426.23605 1276.69361 1275.68633 -0.05 51.09 17/40 Found in E. coli sample 2757 SKVKEAMAPKHK P02666 23.81 4.33 1.00 3 451.92667 1353.76545 1352.75817 -1.29 8.26 22/44 Found in S. aureus sample 3728 YQEPVLGPVRGPFP P02666;E1BIM4 25.84 3.73 1.00 2 778.41815 1555.82903 1554.82175 1.42 59.97 14/26 Found in E. coli sample YQEPVLGPVRGPFP P02666;E1BIM4 32.84 3.07 1.00 2 778.41870 1555.83013 1554.82285 2.13 64.93 12/26 Found in S. aureus sample 3913 HKEMPFPKYPVEP P02666 1.41 3.31 1.00 3 533.60522 1598.80112 1597.79384 -1.57 41.45 16/48 Found in Control sample HKEMPFPKYPVEP P02666;E1BIM4 1.00 4.11 0.20 3 533.60553 1598.80204 1597.79476 -1.00 39.71 17/48 Found in E. coli sample HKEMPFPKYPVEP P02666;E1BIM4 1.00 3.93 1.00 3 533.60577 1598.80277 1597.79549 -0.54 38.86 18/48 Found in S. aureus sample pNpKTpSSKRQVDP A5D7N6 1.00 3.34 3 533.60590 1598.80313 1597.79586 -6.39 39.22 13/52 Found in E. coli sample 4226 YQEPVLGPVRGPFPI P02666;E1BIM4 18.74 3.76 1.00 2 834.95844 1668.90959 1667.90232 -0.77 64.55 15/28 Found in E. coli sample YQEPVLGPVRGPFPI P02666;E1BIM4 4.48 3.16 1.00 2 834.95856 1668.90984 1667.90256 -0.63 66.70 12/28 Found in S. aureus sample LYQEPVLGPVRGPFP P02666;E1BIM4 8.54 3.56 1.00 2 834.95856 1668.90984 1667.90256 -0.63 61.65 13/28 Found in S. aureus sample 4865 RGPGGAWAAKVISNARET Q8SQ28;Q56J78 16.53 3.75 1.00 3 614.33032 1840.97641 1839.96914 -0.73 38.82 21/68 Found in E. coli sample RGPGGAWAAKVISNARET Q8SQ28;Q56J78 28.50 5.18 1.00 4 460.99957 1840.97646 1839.96919 -0.71 38.01 24/102 Found in S. aureus sample 5217 SRQPQSQNPKLPLSILK P80195 31.69 4.53 1.00 4 484.28540 1934.11977 1933.11250 0.79 42.73 26/96 Found in Control sample 5263 QPQSQNPKLPLSILKEK 1.00 3.54 1.00 4 487.78638 1948.12368 1947.11640 0.52 41.88 22/96 Found in Control sample 6017 RGSKASADESLALGKPGKEPR 10.05 3.87 1.00 5 431.63834 2154.16257 2153.15530 -0.05 15.95 27/160 Found in Control sample RGSKASADESLALGKPGKEPR Q4VS17 9.18 3.77 1.00 5 431.63849 2154.16334 2153.15606 0.30 16.34 24/160 Found in Control sample RGSKASADESLALGKPGKEPR 6.92 3.92 1.00 5 431.63867 2154.16425 2153.15698 0.73 15.54 28/160 Found in E. coli sample RGSKASADESLALGKPGKEPR 9.66 4.66 1.00 5 431.63867 2154.16425 2153.15698 0.73 15.71 27/160 Found in S. aureus

Chapter 4, 196

sample 6178 QPQSQNPKLPLSILKEKHL P80195 22.10 4.59 1.00 5 440.45886 2198.26521 2197.25793 -0.20 44.31 24/144 Found in E. coli sample QPQSQNPKLPLSILKEKHL P80195 9.81 4.05 1.00 5 440.45926 2198.26719 2197.25991 0.70 44.51 21/144 Found in Control sample 6465 SSRQPQSQNPKLPLSILKEK P80195 53.89 6.33 1.00 5 456.46353 2278.28855 2277.28127 0.31 34.79 31/152 Found in E. coli sample SSRQPQSQNPKLPLSILKEK P80195 22.40 6.44 1.00 5 456.46353 2278.28855 2277.28127 0.31 35.85 29/152 Found in S. aureus sample SSRQPQSQNPKLPLSILKEK P80195 49.13 5.28 1.00 5 456.46356 2278.28870 2277.28143 0.37 36.94 28/152 Found in Control sample 7095 SRQPQSQNPKLPLSILKEKHL P80195 1.00 3.96 1.00 5 489.08575 2441.39967 2440.39239 0.35 39.30 25/160 Found in Control sample

Table 4-8: The amino acid sequences of potential biomarkers from the LC-MS/MS analysis. Various other parameters are also presented to help better understanding of each of the polypeptides.

Chapter 4, 197

6700

5700 two basic AAs three basic AAs four basic AAs one basic AAs zero basic AAs five basic AAs 4700

3700

2700

1700

700 18 23 28 33 38 43 48 53 58 63 68 Figure 4-7: Milk polypeptides detected in CE-MS analysis based on the number of basic amino acids found in the amino acid sequence. Y-axis represents the molecular mass (kDa) and x-axis represents the CE migration time (min). The red line is specific for the number of basic amino acids in the sequence. Up to five basic amino acids can be found in certain amino acid sequence. The higher number of basic amino acids found in the sequence, the shorter migration time for the CE-MS analysis. This diagram can be used to match the sequence from LC-MS/MS analysis for each polypeptides identified from CE- MS analysis.

Chapter 4, 198

Fq in Fq in Peptide contro Fq in E. Fq in S. ID Sequence Protein Identity l infected coli aureus 2367 PFPEVFGKEKV αS1-casein 0.80 1.00 2757 SKVKEAMAPKHK β-casein 0.50 1.00 3728 YQEPVLGPVRGPFP β-casein 0.30 1.00 3913 HKEMPFPKYPVEP β-casein 0.10 1.00 4226 YQEPVLGPVRGPFPI β-casein 0.20 1.00 4865 RGPGGAWAAKVISN serum amyloid 1.00 0.58 ARET A 5217 SRQPQSQNPKLPLSIL Glycosylation- 1.00 0.07 K dependent cell adhesion molecule 1 5263 QPQSQNPKLPLSILKE Protein 1.00 0.48 K accession not available 6017 RGSKASADESLALGK Fibroblast 1.00 0.59 PGKEPR growth factor binding protein 6178 QPQSQNPKLPLSILKE Glycosylation- 1.00 0.15 KHL dependent cell adhesion molecule 1 6465 SSRQPQSQNPKLPLSI Glycosylation- 1.00 0.74 LKEK dependent cell adhesion molecule 1 7095 SRQPQSQNPKLPLSIL Glycosylation- 1.00 0.07 KEKHL dependent cell adhesion molecule 1

Table 4-9: The protein identities obtained from the UniProtKB database (www.uniprot.org/uniprot). The frequency of each of the identified protein was also presented according to the respected groups.

Chapter 4, 199

4.4 Discussion The mastitic milk samples chosen in this study had significantly higher acute phase proteins concentrations for both SAA and Hp as compared with the healthy milk samples (p<0.05). However, the concentrations for both SAA and Hp were not significant between E. coli and S. aureus infectedgroup (p>0.05). The SCC measurement for all healthy milk samples was less than 100 000 cells/ml which confirmed the healthy status of the cows (average of 46 x 103 cells/ml).

The chosen peptidomic method of capillary electrophoresis (CE) coupled with the electrospray mass spectrometry is known as a rapid (approximately 45 min per sample), sensitive method using its automated approach (Kolch et al. 2005). It was used in this study to detect differences in the milk peptidome between healthy cows (control) and mastitic cows caused by two different bacterial pathogens (E. coli and S. aureus).

All milk samples were analyzed by CE-MS and only polypeptides present at a frequency >70% in at least one of the two groups (control and cases) were chosen for analysis. The compiled data for all test groups (control, all infected, E. coli and S. aureus) are shown in Figure 4-1 and 4-2. It was found that the milk polypeptide patterns of infected cows were markedly different from control cows (Figure 4-1). However, the milk polypeptide patterns of both bacterial groups (E. coli and S. aureus) were more similar (Figure 4-2).

For identification of potential biomarkers for bovine mastitis, milk polypeptides of healthy cows were compared with those of cows with mastitis regardless of its bacterial pathogen. This analysis identified 31 polypeptides of statistical significance after correction for multiple testing (p < 0.05 in max T testing) with 100% sensitivity, specificity and accuracy. The distribution of the polypeptides in these groups is shown in Figure 4-3. It is observed that the identified polypeptides significantly higher in the infected than control group were of lower molecular mass (y-axis) and appeared at much later time of migration (x-axis) which is an indicator of containing a lower proportion of basic amino acids.

To identify biomarkers that could discriminate between two mastitic groups caused by E. coli and S. aureus, the analysis successfully generated three different models with different values of sensitivity, specificity and accuracy. Model 1 appeared to have 100% sensitivity, 91.67% specificity and 96.3% accuracy. Six peaks of discriminatory polypeptides were found in this model. In Model 2, the sensitivity was 100% with the specificity of 83.3% and accuracy of 92.52%. Three peaks of discriminatory polypeptides were found in this

Chapter 4, 200 model. Finally, Model 3 appeared to have 100% sensitivity, 83.3% specificity and 92.52% accuracy. 13 peaks of discriminatory polypeptides were found in this model. In conclusion, a total of 13 discriminatory polypeptides were identified from these three models generated (Table 4-4).

The previous step is known as the learning phase in which the „training set‟ was used to generate a polypeptide pattern with significant biomarkers (Zürbig et al. 2009). To complete biomarker identification exercise, after this training, the model should be able to characterize further unknown data sets so that a correct classification can be confirmed.

There are only a few studies on peptide analysis of milk from cows with bovine mastitis using milk samples (Wedholm et al. 2008; Larsen et al. 2010). Both studies employed the same method in separating the peptides by RP-HPLC with further protein characterization by MALDI-TOF/MS/MS. These studies were also focused on the result of proteolysis caused by different types of proteolytic enzymes. Wedholm et al. (2008) have successfully identified 10 peptides from milk with high SCC (16 000 000 cells/ml) while Larsen et al. (2010) on the other hand identified approximately 20 different peptides in milk samples infused with LTA of S. aureus as a result of proteolysis.

However, the objective of these studies was to gain better understanding in the milk proteinase activities during mastitis and in comparison, this present study is focused on how the peptide profile of the milk samples can be changed in the presence of bovine mastitis. In addition, the discovery of significant peptide biomarker in bovine mastitis was also one of the important objectives in this present study.

Although further model validation will be needed as discussed above, protein identification of these discriminatory polypeptides should be done to provide an initial insight on the change of peptide profiles in milk samples infected with bovine mastitis. It was interesting to see the presence of peptide ID 9094 as one of the discriminatory polypeptides detected from the generated model that can be used to discriminate E. coli from S. aureus infected group (Table 4-3 and 4-4). However, this polypeptide was not detected from the CE-MS analysis that discriminate E. coli from S. aureus infected group (Table 4-6). Although it was not a statistically significant polypeptide that discriminated very well between E. coli and S. aureus infected group, this shouldn‟t be a problem because we were interested in determining the peptide sequencing in the LC-MS system. Thus, it would have been a good idea to be able to identify this polypeptide by sequencing in the LC-MS/MS analysis.

Chapter 4, 201

A total of 45 discriminatory polypeptides (31 polypeptides which discriminate healthy from mastitis cases and 14 polypeptides which discriminate E. coli from S. aureus infected cases) from CE-MS analysis were then analysed for their amino acid sequence by LC- MS/MS As the only parameters used to link the peptides obtained in both systems (CE-MS and LC-MS/MS) were peptide mass and the number of basic amino acids, it can lead to mass ambiguities due to the highly complex peptide mixtures which complicate the assignment of sequences obtained from different separation methods. A further factor was the fact that the mass spectrometers used in the two systems were different. For peptide identification, the LC-MS was used but it was not possible with 100% certainty to correlate the results from the CE-MS with the results from the LC-MS/MS data.

Despite having a problem with mass ambiguity, the results from the LC-MS/MS analysis can be used to determine the protein identity for each of the sequences. However, not all of these 45 polypeptides identified by CE-MS were able to be sequenced by the LC-MS/MS. This occurred as the peptide mass calculated from the CE-MS analysis was not matched with the peptide mass calculated from the LC-MS/MS analysis. It should also be noted that the peptide mass calculated from the LC-MS/MS was based on the mass of peptide plus one proton which made them one unit more than the mass calculated from CE-MS analysis.

Due to the problem in sequencing, a total of 12 polypeptides were able to be sequenced by LC-MS/MS (Table 4-8). From the total of 12 successfully sequenced polypeptides, 11 were used to discriminate control milk samples from the infected milk samples. Only 1 sequenced polypeptide in the set was used to discriminate E. coli infected milk samples from S. aureus infected milk samples. The origin of the milk peptides from which the sequenced polypeptides was found is presented in Table 4-8. Based on their protein accession information, these polypeptides were identified by their amino acid sequences. UniProtKB database was searched for each of the protein identity. Amongst the 12 peptides that were successfully identified were sequences from α S1-casein, β-casein, serum amyloid A protein (SAA), glycosylation-dependent cell adhesion molecule 1 (GlyCAM-1) and fibroblast growth factor binding protein (FGFBP) which were mostly on the bovine protein database as showed in Table 4-9.

As the focus was on discriminatory polypeptides, the discussion will be presented based on the polypeptides/proteins that were able to discriminate between control, infected and two different bacterial pathogens groups. Statistically, α S1-casein, β-casein, GlyCAM-1 and

Chapter 4, 202

FGFBP were the proteins that can be used to discriminate the infected group (regardless of its causing bacterial pathogen) from the control group.

During mastitis, the proteolytic activity in milk can be increased substantially (Verdi et al. 1987). Apart from β-casein, plasmin can also hydrolyses α S2-casein and α S1-casein (Larsen et al. 2006) which explain the presence of peptides of α S1-casein in the infected group higher compared with the control group from the hydrolytic activity in milk.

Statistically, peptides from glyCAM 1 and FGFBP were the peptides that can be used to discriminate the control from the infected groups. The presence of glyCAM 1 protein in bovine milk has been reported previously by Smolenski et al. (2007) and Boehmer et al. (2010). Smolenski et al. (2007) have identified it as one of the minor milk protein that present in all peak lactation, colostrums and mastitic whey and MFGM whilst Boehmer et al. (2010) considered it as one of the host defence proteins. Again, it was also surprising to observe this minor milk protein which acts in the host defence mechanism to appear in the control group in higher frequency compared with the infected group.

The fibroblast growth factor-binding protein (FGFBP) is a secreted carrier protein that releases fibroblast growth factors (FGFs) from the extracellular matrix storage and thus enhancing the FGF activity (Xie et al. 2006). Kawakami et al. (2006) concluded that FGFBP may be tightly associated with lactoferrin in bovine milk. Plath et al. (1998) successfully examined the expression and localization of FGFBP in the bovine mammary gland and it may be important in the local regulation of the bovine mammary gland.

Peptides from SAA (protein ID 4865) were the only discriminatory peptides that were sequenced that were significantly that can be used to differentiate E. coli infected group from S. aureus infected group. Serum amyloid A protein was one of the important acute phase protein in bovine mastitis and it can be produced locally in the mammary gland making it a more reliable indicator of bovine mastitis (Eckersall et al. 2001, 2006) but not a single previous study that able to determine if SAA can significantly discriminate E. coli from S. aureus infected milk samples.

No other identified peptides could be used to discriminate S. aureus infected group from E. coli infected group. Perhaps the use of sample size calculation is needed so that more polypeptides can be sequenced that are able to significantly discriminate between these two groups.

Chapter 4, 203

Although several proteins were successfully identified as the precursor of the peptides using the protein database based on their sequences, those proteins cannot be used per se as the potential markers in each group. The reason is that, it is the peptides of specific fragments of the proteins identified as biomarkers. The protein identity only provided with an insight into which protein were affected and played role in the disease mechanism. It is possible that specific peptides resulted from proteases of the cow rather than bacterial proteases. Thus, no ultimate conclusion can be made based on the protein identity that has been identified in each of the group.

The results discussed above can provide an initial insight on how the milk peptide profiles can change as result of bovine mastitis. However, these results can be improved by applying further improvements. First, validation of SVM models is needed to demonstrate that these milk peptidomes really are able to differentiate healthy cows from mastitic cows with different bacterial-causing pathogens. More milk samples without knowing their infection status (blinded samples) should be tested to validate these models. By validating these models, the status of each marker in each group (control, infected, E. coli or S. aureus infected milk groups) can be confirmed and may eventually be used to determine the presence of bovine mastitis in the cows or even to predict the type of bacterial pathogen that lead to inflammation of the mammary gland.

Second, it was not possible with 100% certainty to correlate the results from the CE-MS with the results from the LC-MS/MS data based on peptide masses, this situation can lead to mass ambiguities due to the highly complex peptide mixtures involved which complicates the assignment of sequences obtained from different separation methods and also due to the fact that the mass spectrometers used after CE-MS and LC-MS/MS were different. Therefore, further work should be conducted in order to resolve this ambiguity issue by running samples with internal standards of known mass in both systems. In this way, an internal mass calibration of each sample can be done that would allow a more correct cross-correlation of results between both systems.

It would also be interesting to propose a time-course of infection study for future work and determine the earliest time that can be applied to diagnose an infected cow based on the proposed polypeptide biomarker of bovine mastitis and if the putative biomarkers would be able to discriminate subclinical mastitis from clinical mastitis and healthy group. More information on peptide sequencing can lead to a better understanding of the

Chapter 4, 204 pathophysiology of bovine mastitis and eventually could help in improving the bovine mastitis detection.

Chapter 5, 205

Chapter V: Metabolomics Analysis in Bovine Mastitis

5.1 General Metabolomics Being one of the most newly emerged “-omics” sciences, metabolomics is becoming one of the most significant field of study in system biology which complements the other “- omics” and proving to have unique advantages. It was first proposed by Oliver (1998), and seeks to characterize and quantify the whole collection of both intra and extra-cellular metabolites. The metabolome is defined as the qualitative and quantitative collection of all low molecular weight of molecules that are present in biological fluid, organism or cell under a different set of physiological conditions (Tyagi et al. 2010; Banderas et al. 2011). These metabolites include a diverse range of endogenous and exogenous chemical molecules such as peptides, amino acids, nucleic acids, carbohydrates, organic acids, vitamins, polyphenols, alkaloids and any other chemical that is used, ingested or synthesized by a given cell or organism (Wishart 2008). Some of the metabolites are acquired externally but most are products or intermediates from metabolic reactions (Tomita and Nishioka 2005).

Metabolomic analyses involve two basic methodologies (Nielsen and Jewett 2007). First is metabolite profiling which allows qualitative screening of a large number of metabolites (usually more than 100) and the pattern of metabolites produced via high-throughput detection can be used to find discriminatory elements followed by data deconvolution methods. This method covers metabolic fingerprinting (analysis of endometabolome) and metabolic footprinting (analysis of exometabolome). According to Álvarez-Sáchez et al. (2010), global metabolomics profiling involved detection of metabolites by using a single analytical platform or a combination of complementary analytical platforms (GC-MS, LC- MS, CE-MS or NMR). The second method is target analysis which aims to produce absolute or semi-quantitative detection of metabolites. Previously, only small numbers of metabolites were able to be analyzed but with new development on quantitative analysis, more expanded metabolite coverage can be performed.

However, the comprehensive analysis of a wide range of metabolites is only possible with the use of recent technological advancements in separating and identifying these molecules (Wishart 2008). The most widespread analytical platforms include robust, high-resolution mass spectrometry (MS) instruments, nuclear magnetic resonance (NMR) spectrometers, Chapter 5, 206 capillary electrophoresis (CE) and ultra-high pressure liquid chromatography (UPLC and HPLC). Many investigations have used an integration of various analytical methods such as gas chromatography combined with mass spectrometry (GC-MS) and liquid chromatography combined with mass spectrometry (LC-MS) to improve metabolite identification and coverage (Álvarez-Sáchez et al. 2010).

There are many advantages to studying the metabolome in comparison with transcriptome and proteome. According to Dunn & Ellis (2005) and Álvarez-Sáchez (2010) metabolomics is often seen more useful because the number of metabolites is smaller than the number of proteins and genes in a cell which reduces the sample complexity. Furthermore, the concentration of metabolites can significantly change during biochemical reaction even though the concentration of enzyme and metabolic flux may not be significantly altered (Kell and Mendes 1999). As the downstream product of gene expression, the metabolome reflects the functional level of the cell more appropriately and is thought to be amplified in physiological changes as compared to proteome and transcriptome. Furthermore, metabolic fluxes are more influenced by environmental stress than gene expression, making the measurement of metabolites more appropriate. Finally, metabolic profiling is less expensive than proteomic and transcriptomic experiments.

Despite having many benefits for the analysis of smaller molecules, one major problem in metabolomics is the lack of standardized methods especially in the analysis of global metabolomics (Álvarez-Sáchez et al. 2010). A standard metabolomic experimental approach is needed from sample collection through chemical analysis to data processing.

Metabolomics has been applied to various investigations such as determining the metabolic biomarkers that act as indicators to the presence of a disease or as a response towards drug- based intervention. Other than that, various metabolomic studies have been performed on plants and microbes (Le Gall et al. 2003) and on bacterial characterisation (Vaidyanathan et al. 2002). For the past few years, metabolomic analyses has gained an interest in food and nutrition fields to give a more comprehensive molecular picture on food composition as well as food consumption.(Wishart 2008).

5.1.1 Metabolites in milk and serum found during bovine mastitis Metabolomics provide a very important tool to analyse changes in milk metabolome during the presence of bovine mastitis because of its rapid and high throughput characterization of small molecule metabolites found in milk or serum. Chapter 5, 207

There have been many studies on the metabolites in milk and serum in relation to mastitis from the era when metabolism study used traditional methodologies. An early study by Peter et al. (1990) investigated the metabolites of prostanoids in milk of heifers after intramammary infusion of E. coli organisms. Prostanoids are known to be important mediators in the inflammatory processes associated with endotoxin-induced mastitis. Increases of arichidonic acid metabolites such as thromboxane A2, prostaglandin E2 (PGE2) as well as milk prostacyclin (PCM) were seen in milk from infected quarters. The increased concentrations of these metabolites showed that they have an important role in the pathophysiologic process of coliform mastitis released as the mediators during inflammation.

The presence of change in metabolites during mastitis is also associated with other factor such as dietary selenium (Se). A study by Maddox et al. (1991) showed that cows with Se deficiency (-Se) contained significantly higher concentrations of PGE2, 6-keto- prostaglandin F1α and leukotriene B 4 (LTB4) in milk after E. coli inoculation as compared with milk from cows with Se sufficiency (+Se). The higher eicosanoid concentration in milk was associated with higher number of bacteria. This is may be due to impaired killing abilities of selenium deficiency polymorphonuclear leukocytes (PMNL) as during non-specific immune response such as in coliform mastitis, the function and activity of PMNL are highly influenced by eicosanoids. Greater concentration of these eicosanoid levels in milk of –Se cows is due to the increase of substrate availability and enzyme activity from the activation of cyclooxygenase and lipoxygenase enzymes in the Se- glutathione peroxidase (Se-GSH-Px) depleted state. The severity of clinical signs seen in – Se cows could be at least partially caused by the higher eicosanoid concentrations in the milk as they have play an important role in the pathogenesis of inflammatory and endotoxin-induced diseases.

Serum trace minerals were also studied (Erskine and Bartlett 1993) during E. coli-induced mastitis in which a transient depression in both plasma zinc (Zn) and iron (Fe) was seen during intramammary challenge with E. coli. High number of bacteria (E. coli) secreted in milk over time was correlated with the severity of clinical signs of infection as well as with decreased plasma Zn and Fe concentrations. These divalent cations are important in the growth of gram-negative bacteria such as E. coli and reduced concentrations of both Zn and Fe may be part of defence mechanism to limit the bacterial growth. As for copper (Cu) concentration, this study did not show any increase during the inflammatory process but in fact, it decreased to 75% of pre-challenge levels 24 hours post inoculation. This is not in Chapter 5, 208 accordance with the previous study by Conner et al. (1986) in which they found increased ceruloplasmin during the acute phase response in cattle as well as studies by Barber et al. (1988) and Etzel et al. (1982) where they found increased in serum ceruloplasmin in both rats and hamster. Ceruloplasmin is the Cu-binding protein which plays a role as a scavenger for superoxide anion radicals. The difference might be due to species differences.

Gram negative mastitis in early lactation may also influenced the ovarian and certain endocrine functions and metabolism in dairy cows. Huszenicza et al. (1998) found that during lipopolysaccharide (LPS) loading, various events occurred that might impair the ovarian functions such as delay in the resumption of cyclicity or even causing premature luteolysis if sensitive corpus luteum is present simultaneously on the ovary caused by prostaglandin F2α. However, based on individual progesterone (P4) profiles in the blood of the affected cows, gram negative mastitis influenced the ovarian function during the 3rd up to 10th weeks of lactation. During early puerperal (<14 days) mastitis, its prevalence was not influenced by the concentration of acetoacetone, β- OH butyrate (BHB) or nonesterified fatty acids. Cows affected with different forms of early puerperal mastitis did not show any delay in the onset of cyclicity either.

Interesting metabolites that exist in the milk are the reactive oxygen metabolites that - include superoxide anion (O2 ), hydroxyl radical (OH), hydrogen peroxide (H2O2) and fatty acid radicals (Lindmark-Månsson and Åkesson 2000). These reactive metabolites are released by PMN after phagocytising foreign particles or bacteria and discharging their granules’ contents. Although they are harmful to bacteria, these oxidants are known to be cytotoxic and can induce tissue damage by destroying host cells. To control these reactive oxygen radicals, anti-oxidants are also present in the milk during the inflammatory processes which include natural vitamins (vitamin E and C), beta-carotene and enzymatic systems mainly represented by superoxide dismutase (SOD), catalase and glutathione peroxidase (GSHPx). A study by Hamed et al. (2008) investigated the level of these anti- oxidants in association with the bulk milk somatic cell counts (BMSCC). Because oxidants released by PMN induce cellular damage (Weiss 1989; Van Asbeck 1990) and antioxidants are needed to control these reactive oxygen metabolites, the relationship between milk somatic cells and antioxidant activities was studied. SOD activity was found to be unaffected by BMSCC. SOD activity was weakly dependent upon BMSCC and suggested that SOD incorporation did not prevent the cytotoxic effect of the oxygenated radicals and pointed out the importance of other anti-oxidants. As for the catalase, it showed increased Chapter 5, 209 activity in high BMSCC samples probably playing a central role in milk redox control especially during mastitis. It also could be used as an indicator in mastitis. This study also showed a positive correlation between GHSPx activity with the BMSCC especially with the neutrophil content but its role in bovine milk was thought to be limited. However, it was concluded that catalase and GSHPx are the principal enzymatic defence that neutralize the harmful free radicals.

Metabolic activity is able to influence the health of the udder. Near parturition and in the period of early lactation, there is a large nutrient demand in which nutrients are mobilized from tissue reserves. This results in changes in blood concentration of both hormones and metabolites (Ingvartsen and Friggens, 2005). In general, primiparous cows have a greater prevalence of mastitis as several important immune factors are impaired around calving. Thus, one study by Nyman et al. (2008) revealed that blood concentration of β- hydroxybutyrate (BHBA), nonesterified fatty acids (NEFA), glucose and α-tocopherol around calving were associated with the SCC at first test-milking of primiparous cows. Immune variables such as the collectins which consist of collectin-43 (CL-43) and conglutinin as well as haptoglobin (Hp) were also seen to be correlated with Box-Cox transformed SCC (bcSCC) and these correlations must be based from a direct breed effect. It is also known that age and breed factors were significantly associated with several metabolites and immunological variables. For example, heifers calving at an older age had greater concentrations of serum BHBA and NEFA but lower glucose, insulin and urea nitrogen values compared with heifers calving at a younger age (Nyman et al. 2008). This has shown that older heifers were in more severe negative energy balance in the period around calving and in early lactation than younger heifers.

Another study that investigated various metabolites in serum and their variation during the transition period in dairy cows was performed by Moyes et al. (2009). This study is in agreement with Nyman et al. (2008) in which there was an increased NEFA concentration in cows that developed clinical mastitis at calving and during early lactation. Thus, higher circulating NEFA was shown to inhibit immune responses that lead to the development of mastitis especially during early lactation. Clinical mastitic cows also showed higher circulating BHBA concentrations during the transition period. This result was in contrast with that of Nyman et al. (2008) in which high BHBA concentrations were associated with lower bcSCC at the first test-milking during early lactation. According to Nyman et al. (2008), this increase may be associated with microbial butyrate production from rumen carbohydrate fermentation, which is then converted to BHBA in the rumen epithelium. As Chapter 5, 210 for glucose levels, the results were inconclusive. Higher glucose levels were seen during pre-partum period in cows that developed clinical mastitis as compared to cows that did not developed mastitis during early lactation period. This can be explained by the lack of difference in insulin concentration in clinical mastitic cows and this might suggested a lower efficiency of glucose clearance from circulation for these cows compared with subclinical mastitic and healthy cows. This study also suggested that aspartate aminotransferase (ASAT) has no association with the development of clinical and subclinical mastitis during the first week of lactation making it a less useful marker to evaluate the risk of mastitis due to low specificity of ASAT in lactation. Overall, NEFA was the most consistent with least conflicting positive association with the development of mastitis and making it the most useful marker for the risk of mastitis during early lactation. The use of metabolites as the risk factors of mastitis is promising but further investigations are needed.

Melendez et al. (2009) studied a relationship between NEFA at calving with the incidence of many periparturient diseases including mastitis in Holstein dairy cattle. The authors agreed that increased in NEFA concentrations in serum have been associated with impaired immunity which predisposed cows to periparturient diseases. Thus the NEFA values at calving can be used to monitor the nutritional management as well as to be used as indicators in the risk factors in many periparturient diseases including mastitis.

Metabolites can also be produced by the mastitis-causing pathogens. These metabolites have been used to determine the pathogen. This is based on the fact that all microorganisms produced their own range of often volatile metabolites. Hettinga et al. (2008) identified a wide range of volatile metabolites that were produced by different types of microorganisms using a solid-phase microextraction (SPME) coupled to gas chromatography/mass spectrometry (GC/MS). In this study, they found that volatile metabolites found in milk from uninfected quarters differed significantly from those found in milk from infected quarters. Lower numbers and amounts of metabolites were found in the samples from uninfected quarters compared to the samples from mastitis quarters. For example, S. aureus and E. coli both formed a large amount of ethyl acetate and acetic acid. It was confirmed that different microorganisms produced different types of metabolites which means that the presence of certain volatile metabolites in infected milk samples could be used to detect and identify mastitis-causing pathogens.

A further study on volatile metabolites in mastitis milk samples by the same authors investigated the origin of these components (Hettinga et al. 2009). Although it is known Chapter 5, 211 that pathogens are responsible to the presence of volatile metabolites in mastitic milk samples, some of these detected metabolites may also be formed by enzymes, present in the blood of the animals but which travel into the milk due to the disruption of blood-milk barrier during inflammatory process. Around 20% of metabolites identified in this study were found to be different between inoculated milk (milk from healthy animals inoculated with the same bacteria) and mastitic milk samples. Nevertheless, most of the volatile metabolites detected in mastitis samples are mainly formed by the pathogens except for ethyl esters of free fatty acids (FFA) which were absent or significantly lower in inoculated samples. These metabolites are formed by the esterase enzymes found in cow’s blood. During mastitis, due to increase leaking of the barrier between blood and udder, esterase enzymes can be found in the mastitic milk (Raulo et al. 2002). Thus, during mastitis, this enzyme can probably transfer from the blood to the milk due to disruption of the blood- milk barrier. Furthermore, dimethyl sulphide may be useful for E. coli identification as it was found that dimethyl sulphide concentrations is increased in samples inoculated with E. coli.

Another type of metabolite which is seen to have an effect in milk production are the metabolites of nitric oxide (NO). Bouchard et al. (1999) have investigated the nitric oxide production during endotoxin-LPS induced bovine mastitis. In this study, a significant three- to five-fold of increase in nitrate and nirite contents (metabolites of nitric oxide) were formed in milk samples upon LPS infusion. Nitric oxide is produced from L-arginine by the enzymes termed as nitric oxide synthase (NOS). Inducible isoforms of NOS (iNOS) are also thought to be cause overproduction of nitric oxide. It is known that somatic cells are responsible for the production of nitric oxide although in this study, it did not establish of which cell type(s) that is (are) involved. They have also studied the effect of co-infusion of LPS with an inhibitor of iNOS; aminoguanidine which successfully prevented the nitric oxide production upon LPS infusion. These results confirmed the activity of iNOS during the endotoxin-induced inflammatory reaction.

During lactation, nitric oxide synthase plays a role in generating nitric oxide contained in the mammary gland tissues (Lacasse et al. 1996). Nitrate and nitrite are the end products of nitric oxide oxidation in aqueous solutions (Moncada et al. 1991). Milk also contains S- nitrosothiols (RSNO) which are also metabolites of short-lived NO. Because milk contains lactoperoxidase, nitrite can be oxidized to toxic nitrogen dioxide (Silanikove et al. 2005). A study by Titov et al. (2010) revealed by using an enzyme sensor method that in healthy milk and colostrums, no nitrite compounds were found. Thus, its presence in milk can be Chapter 5, 212 an indicator of mastitis. However, the occurrence of these compounds was accompanied with a lower content of RSNO which possibly occurs under the effect of reactive oxygen species. Nitrite also can be produced by leukocytes. As said earlier, the oxidation of nitrite by lactoperoxidase can lead to the production of toxic nitrogen dioxide, a powerful oxidant that could cause the oxidative breakdown of milk lipoprotein. To avoid this toxic effect, the presence of high concentrations of other compounds that are able to compete with nitrite for lactoperoxidase is important (thiocyanate and ascorbate).

5.1.2 Aims of Study In conclusion, the aim of this study is to apply metabolomic study by the use of LC- MS/MS analysis to study the presence of important metabolites that can be used to discriminate normal milk samples from milk samples from quarters infected with bacteria causing bovine mastitis. It was hoped that by using this sensitive and high-throughput methodology, significant results could be achieved and lead to the discovery of important and significant mastitis biomarkers and that differences may be found in samples of milk from infected quarters caused by different pathogens.

5.2 Materials and Methods 5.2.1 Animals and Samples

Milk samples from clinical mastitic cases were sent from Vale Veterinary Laboratory, Devon, UK and bacteriological culture was performed in the Bacteriology Laboratory of Vale Veterinary Laboratory to determine the bacterial cause of bovine mastitis. Milk samples from mastitis caused with either E. coli (n=10) or S. aureus (n=6) infection were collected for further analysis. Control milk samples (n=10) from healthy cows from the Cochno Farm of School of Veterinary Medicine, University of Glasgow were also used in this study as comparison to those clinical mastitic cases. The healthy milk samples were confirmed by somatic cell counts (<100 000 cells/ml). Prior to metabolomic analysis, SAA and Hp assays (Chapter 2) were performed determine the concentration of both acute phase proteins in each milk sample for further confirmation of disease status. Samples were stored at -20°C prior to analysis.

5.2.2 Chloroform and methanol Extraction

5.2.2.1 Reagent preparation A mixture (1:3) of chloroform: methanol was prepared by adding 20ml of chloroform into 60ml methanol. Chapter 5, 213

5.2.2.2 Extraction method An aliquot of 500µl of each milk sample was added to 2ml of 1:3 chloroform: methanol solution and mixed vigorously in the cold room at 4ºC for 2 hours. Next, the sample was centrifuged at 13 000 x g for 3 minutes at 4ºC. The supernatant was aspirated and was then stored at -80ºC until analysis by LC-MS.

5.2.3 LC-MS Analysis

LC/MS was carried out using a Dionex UltiMate 1 coupled to an Orbitrap Exactive. The separation of metabolites was performed using a 4.6 x 150mm ZIC-HILIC column at 300μl/min. The gradient ran from 80% acetonitrile (ACN) to 20% ACN in 30 minutes, with a step to 5% ACN for 9 minutes. Re-equilibration was then performed for 6 minutes at 80% ACN. 10μl of each sample was used per run. The MS acquisition was performed in both negative and positive ionisation modes with full scan modes. The MS was set at 50,000 resolution with the scan range from 70-1400 amu. The LC-MS analysis was performed under supervision of Dr. Karl Burgess of Sir Henry Wellcome Functional Genomic Functional, University of Glasgow, UK.

5.2.4 Data Acquisition

The LC/MS data was analyzed using MzMine software (version 2.0) (http://mzmine.sourceforge.net/index.shtml; Finland and Japan). For the peak detection, a noise threshold of 5000 and m/z tolerance of 0.0005 Da was performed. An alignment with a mass tolerance of 0.0005 and a retention time tolerance was 15 seconds.

5.2.5 Analysis of Individual Metabolites

Output from the LC/MS for both positive and negative modes was obtained as spreadsheet of mass-to-charge ratio peaks for each sample and with m/z values providing molecular formula by comparison to metabolite database and a putative identification was based on the formula. Not all m/z peaks gave identified metabolites. For each peak, the mean and SD of the m/z intensities for all samples in the group (control; E. coli; S. aureus) were automatically determined. Inspection of the results and manipulation of the spreadsheet allowed analysis in a variety of criteria.

For both positive and negative modes, the metabolites were sorted to identify those with the 20 metabolites with highest m/z intensities in each of the 3 groups. Chapter 5, 214

Numerous metabolites were not present in the control group but were present in either or both the E. coli and S. aureus groups. These were identified and the 20 metabolites with largest difference in mean m/z intensities determined by substraction.

Numerous metabolites had large difference between the mean of control group and the mean of either E. coli or S. aureus groups. They were determined by calculation of the total increase or decrease by dividing the mean of the m/z intensities of the infected groups by the mean of the control group and the 15 metabolites with greatest fold increase or decrease were identified.

Metabolites showing significant difference between the E. coli and S. aureus groups were identified by calculation of the fold change by dividing the median of the m/z intensities of the S. aureus group by the median m/z intensities of the E. coli group. The 15 metabolites with the highest fold change between groups increased in either S. aureus or in E. coli were identified and the Mann Whitney test for significant differences applied for non-parametric data. The metabolites were listed in order of increasing probability of the significance of a difference (P-value).

5.2.6 Metabolites Analysis – Pathway Analysis

Bioinformatic analysis of the metabolomics data performed using tools provided by the Scottish Metabolomics Facility. An in-house web facility, known as PATHOS (http://motif.gla.ac.uk/Pathos/index.html), has the ability to display metabolites identified based on the metabolic pathways in which they occur. By listing areas of metabolism and by generating a Kyoto Encyclopaedia of Genes and Genomes (KEGG) map, the data of experimental m/z values or refined lists of metabolites generated by the MS data by other means was analysed. PATHOS also provides quantitative data for the metabolites identified and these changes are indicated by colour-coding on KEGG map.

For preliminary analyses, changes in metabolites were grouped based on different metabolism; lipid, protein and carbohydrate metabolisms. In each group, the analysis was focussed on six metabolic pathways showing most changes for both positive and negative ion modes. Chapter 5, 215

5.3 Results

5.3.1 SCC measurement and acute phase proteins concentration In order to confirm the disease status of milk samples used in this study, SCC measurement and acute phase proteins (SAA and Hp) concentrations were determined. Table 5-1 showed the SCC measurement and APP concentrations in all control milk samples used in this study. The results were as the same as in Chapter 4, Section 4.3.1. Table 5-2 (a and b) showed the bacteriological culture results and the concentration of both SAA and Hp for mastitic milk samples caused by E. coli and S. aureus. The results were also as the same as in Chapter 4, Section 4.3.1. The Statistical analysis (Mann Whitney) showed that both SAA and Hp were significantly higher (p=0.00; p<0.05) in the infected group (E. coli and S. aureus) but not significantly different between two bacterial groups (p=0.958; p>0.05 for SAA and p=0.428; p>0.05 for Hp). Chapter 5, 216

Control milk samples

Table 5-1: The acute phase proteins (SAA and Hp) concentrations and SCC measurement in all control milk samples used in this study. The range of SAA concentrations was between 0.08 and 2.96 μg/ml with an average of 0.71μg/ml and a median of 0.48μg/ml. The range of Hp concentrations was between 1.95 and 32μg/ml with an average of 12.8μg/ml and a median of 12.64μg/ml. The range of SCC measurement was between 10 and 96 x 103 cells/ml with an average of 45 x 103 cells/ml and a median of 46 x 103 cells/ml. Chapter 5, 217

Table 5-2a: The acute phase proteins (SAA and Hp) concentrations and bacterial causing mastitis in infected milk samples caused by E. coli used in this study. The range of SAA concentrations was between 4.18 and 1500μg/ml with an average 626.6μg/ml and a median of 90.79μg/ml. The range of Hp concentrations was between 16.03 and 354.26μg/ml with an average of 112.5μg/ml and a median of 98.85μg/ml. Chapter 5, 218

Table 5-2b: The acute phase proteins (SAA and Hp) concentrations and bacterial causing mastitis in infected milk samples caused by S. aureus used in this study. The range of SAA concentrations was between 15.99 and 1500μg/ml with an average of 526.1μg/ml and a median of 51.54μg/ml. The range of Hp concentrations was between 22.15 and 111.8μg/ml with an average of 71.44μg/ml and a median of 78.49μg/ml. Chapter 5, 219

5.3.2 Metabolites found in milk Upon LC-MS analysis, the identified metabolites were analyzed by MzMine software, the data was presented based on the individual metabolites found according to their m/z values, retention times, numbers of detected peaks, molecular names and their abundances in each of the milk samples. The data was divided into positive and negative ionisation modes based on the MS acquisition. A total of 9904 were detected in the milk samples using positive mode detection of which 814 were identified using the MzMine software. For the negative ion detection the equivalent figures were a total of 5551 metabolites detected with 512 being identified. Numerous metabolites were present with both ion detection systems and inspection of the data showed that 221 metabolites were in both the positive and negative mode so that a total 1105 metabolites were detected across the milk samples analysed (Table 5-3). The mean values of m/z intensities of metabolites in each group were then determined. The full excel files with the identified and non-identified metabolites are presented in the supplementary information. The following examination of the metabolites and their variation between milk from healthy cows and that from udder quarters with mastitis caused by E. coli or S. aureus are focused on the metabolites that were identified in the samples by LC-MS-MS and the MzMine software programme.

With the large data set generated by the study, the results were examined by scrutiny of the individual metabolites and their variation in mean m/z intensity between control and infected groups. In order to reduce the complexity of the interpretation, the metabolites are discussed in relation to the m/z intensity and the mean of the peaks obtained with the discussion focused on the metabolites with the highest intensity or on those with the greatest relative value between bacterially infected groups. Thereafter the major possible metabolic pathway alterations that produce the identified changes in metabolites are described. Chapter 5, 220

Positive Negative Total of Mode Mode metabolites Number of metabolites 9904 5551 15455 Number of identified metabolites 814 512 1326 Number of unidentified metabolites 9090 5039 14129 Number of common identified 221 metabolites Number of individual identified 1105 metabolites

Table 5-3: The total number of metabolites, identified and unidentified in both positive and negative ionisation modes. The number of common and individual identified metabolites was also presented. A total of 15 455 metabolites were identified from the LC-MS/MS analysis. However, only 1326 metabolites were successfully identified by the MzMine software. About 221 of common metabolites and 1105 individual metabolites were identified from this software. Chapter 5, 221

5.3.3 Effect of mastitis on the metabolites of milk The first 20 metabolites with highest m/z intensities in each three groups of milk samples; control, E. coli and S. aureus groups were tabulated in each positive and negative ionisation modes as presented in Table 5-4 and 5-5. From the positive mode MS, it was found that the metabolites with greatest intensity differed between the groups although most of the highest metabolites in the control were still within the top group. The 3 metabolites with highest intensity in the control group (Table 5-6) were L-acetylcarnitine, choline and sn-glycero-3-phosphocholine. The fourth metabolite was lactose, the primary carbohydrate of milk. However, the Pathos analysis programme identified this as 1-alpha- D-Galactosyl-myo-inositol which has an identical molecular formula of C12H22O11. On tables, the form given is maintained with lactose identified in bracket. In the group with E. coli infection (Table 5-7) choline was the metabolite with highest m/z intensity but L- carnitine and the amino acid, L-valine were also at the top of the metabolites detected while lactose had the fourth highest intensity. In the group with S. aureus infection (Table 5-8), L-valine had the highest intensity and it was noticeable that 11 of the top 20 were amino acids, di-peptides or tri-peptides, compared to 6 in the E. coli group and only one, glutamate in the milk samples from the control group. Lactose was detected in the S. aureus group but was tenth of the twenty metabolites with the highest m/z intensity.

With the negative ion detection, all three groups had the same metabolites with the two highest m/z intensities, being citrate (though incorrectly identified as 5-dehydro-4deoxy-D- glucarate) and lactose (Tables 5-9 to 5-11), which reflect their high concentration in milk. In the samples for both the E. coli and S. aureus infections the third highest m/z intensity was lactate, which was not in the top 20 in the control group and its presence in the samples from mastitic milk would indicate the alteration in metabolism caused by the bacteria. Also noticeable was that that glucose/galactose (identified as myo-inositol) was detectable in milk from the control group and that from the E. coli infected but not the S. aureus infected groups. It was noticeable that the metabolites in the top 20 of the negative ion detection were mostly small molecular weight acidic metabolites while those in the positive ion detection had a higher proportion of basic metabolites.

Metabolites which were absent in control milk samples but present in both infected groups (E. coli and S. aureus) were examined in both positive and negative ion modes. Twenty metabolites with the largest difference between intensities in E. coli and S. aureus infected groups were tabulated in Table 5-12, 5-13, 5-14, 5-15. From the positive mode analysis, of the 20 metabolites which were higher in the S. aureus group than E. coli 18 were Chapter 5, 222 tripeptides, with Leu-Val-Ser being the metabolite with greatest difference from that in the E. coli group (Table 5-12). In contrast ten different tripeptides had higher intensities in the samples from the E. coli group compared to the S. aureus group (Table 5-13). However also detected in this group were other metabolites such as calligonine, harmolol (both are alkaloids) and lincomycin. From the negative mode analysis, of the 20 metabolites which were higher in the S. aureus group than E. coli, 12 were tripeptides with Leu-Gln-Ser being the metabolites with greatest difference from that in the E. coli group (Table 5-14). In contrast, eight different tripeptides had higher intensities in the samples from the E. coli group compared to the S. aureus group (Table 5-15). Chapter 5, 223

S. Name Control Name E. coli Name aureus L-Acetylcarnitine 5.3E+08 Choline 6.9E+08 L-Valine 4.7E+08 Choline 4.6E+08 L-Carnitine 3.3E+08 Choline 4.6E+08 sn-glycero-3-Phosphocholine 4.6E+08 L-Valine 3.2E+08 L-Phenylalanine 4.2E+08 1-alpha-D-Galactosyl-myo-inositol 1-alpha-D-Galactosyl-myo-inositol (lactose) 4.1E+08 (lactose) 3E+08 L-Acetylcarnitine 3.5E+08 L-Carnitine 4E+08 L-Acetylcarnitine 3E+08 Piperidine 3.2E+08 [FA methyl(4:2/18:2)] methyl 9- butylperoxy-10_12-octadecadienoate 2.3E+08 L-Phenylalanine 2.2E+08 Leucyl-leucine 3E+08 [ST] (5Z_7E)-9_10-seco-5_7_10(19)- Methyl acetyl ricinoleate 1.9E+08 cholestatriene 2.1E+08 L-Carnitine 2.8E+08 [ST] (5Z_7E)-9_10-seco- 5_7_10(19)- Methacholine 1.7E+08 Piperidine 1.9E+08 cholestatriene 2.7E+08 [FA methyl(4:2/18:2)] methyl 9- L-Glutamate 1.5E+08 butylperoxy-10_12-octadecadienoate 1.8E+08 L-Proline 2.5E+08 1-alpha-D-Galactosyl- N-Acetyl-D-glucosamine 1.5E+08 Methyl acetyl ricinoleate 1.6E+08 myo-inositol (lactose) 2.5E+08 Glycogen 1.5E+08 Methacholine 1.4E+08 L-Glutamate 2.4E+08 [ST] (5Z_7E)-9_10-seco-5_7_10(19)- sn-glycero-3- cholestatriene 1.4E+08 Glycogen 1.3E+08 Phosphocholine 2.3E+08 1-O-(1Z-Tetradecenyl)-2-(9Z- octadecenoyl)-sn-glycerol 1.2E+08 L-Tyrosine 1.2E+08 L-Tyrosine 2.1E+08 [FA oxo(19:0)] 10-oxo-nonadecanoic [FA oxo(19:0)] 10-oxo- acid 1.1E+08 Leu-Tyr 1.1E+08 nonadecanoic acid 1.6E+08 Chapter 5, 224

[FA methoxy_hydroxy(18:2)] 8- methoxy-13-hydroxy-9_11- octadecadienoic acid 1.1E+08 L-Proline 1.1E+08 Leu-Phe 1.5E+08 [FA oxo(19:0)] 10-oxo-nonadecanoic O-Propanoylcarnitine 9.3E+07 acid 1.1E+08 Leu-Tyr 1.5E+08 [FA (17:1/2:0)] 8E-Heptadecenedioic acid 8.8E+07 L-Glutamate 1E+08 Leu-Val-Ser 1.5E+08 [FA methoxy_hydroxy(18:2)] 8- methoxy-13-hydroxy-9_11- Mercaptoethanol 7.6E+07 octadecadienoic acid 9.7E+07 Leu-Leu-Val 1.4E+08 1_3_8-Naphthalenertriol 7E+07 sn-glycero-3-Phosphocholine 8.7E+07 Leu-Thr 1.3E+08 [FA (17:1/2:0)] 8E-Heptadecenedioic O-Butanoylcarnitine 6.6E+07 acid 7.8E+07 Methacholine 1.3E+08

Chapter 5, 225

Name Control Name E. coli Name S. aureus 5-Dehydro-4-deoxy-D-glucarate (citrate) 1.8E+09 5-Dehydro-4-deoxy-D-glucarate (citrate) 1.4E+09 5-Dehydro-4-deoxy-D-glucarate (citrate) 1.7E+09 1-alpha-D-Galactosyl-myo-inositol (lactate) 1.5E+09 1-alpha-D-Galactosyl-myo-inositol (lactate) 1.2E+09 1-alpha-D-Galactosyl-myo-inositol (lactose) 1.2E+09 N-Acetyl-D-glucosamine 6-phosphate 2.6E+08 (R)-Lactate 8.1E+08 (R)-Lactate 9.2E+08 cis-Aconitate 1.7E+08 myo-Inositol (glucose/galactose) 1.8E+08 L-Glutamate 1.6E+08 myo-Inositol (glucose/galactose) 1.1E+08 Succinate 1.8E+08 cis-Aconitate 1.2E+08 Hippurate 9.9E+07 cis-Aconitate 1.1E+08 myo-Inositol (glucose/galactose) 1.2E+08 Urate 9E+07 2-C-Methyl-D-erythritol 4-phosphate 9.2E+07 (S)-Malate 1E+08 2_5-Dioxopentanoate 8.4E+07 2-Dehydro-3-deoxy-L-arabinonate 8.6E+07 Urate 9.4E+07 Miserotoxin 7.9E+07 Pyruvate 8.2E+07 2-C-Methyl-D-erythritol 4-phosphate 8.5E+07 2-C-Methyl-D-erythritol 4-phosphate 7.8E+07 Creatine 7.3E+07 L-Erythrulose 7.5E+07 L-Erythrulose 7.5E+07 Hippurate 7.2E+07 N-Acetyl-D-glucosamine 7.4E+07 N-Acetyl-D-glucosamine 7.4E+07 (S)-Malate 7.1E+07 Miserotoxin 7.1E+07 Creatine 7.4E+07 2-C-Methyl-D-erythritol 4-phosphate 6.3E+07 2_5-Dioxopentanoate 6.9E+07 (S)-Malate 7.4E+07 Urate 6.1E+07 Taurine 6.2E+07 L-Glutamate 7.4E+07 L-Glutamate 6E+07 (R)-3-Hydroxybutanoate 6.2E+07 sn-glycero-3-Phospho-1-inositol 6.6E+07 2_5-Dioxopentanoate 6E+07 Pyruvate 5.8E+07 D-Glucose 6-phosphate 5.2E+07 N-Acetyl-D-glucosamine 6-phosphate 5.9E+07 Creatine 5.7E+07 2-Dehydro-3-deoxy-L-arabinonate 4.4E+07 Glycerol 2-phosphate 5.2E+07 Hippurate 5.4E+07 2-Butyne-1_4-diol 4.4E+07 2-Oxoglutarate 4.8E+07 2-Dehydro-3-deoxy-L-arabinonate 4.5E+07 Pantothenate 4.3E+07 (2S)-2-Isopropylmalate 4.6E+07 Mesaconate 4.3E+07 Table 5-5 : The first 20 of metabolites with the highest m/z intensity of negative ionisation mode from the total of 512 metabolites identified using the MzMine software in control, E. coli infected and S. aureus infected groups. Chapter 5, 226

Controls row m/z Molecular formula Name m/z intensity 204.12288 C9H17NO4 L-Acetylcarnitine 532069651.6 104.10717 C5H14NO Choline 459631082.1 258.10994 C8H21NO6P sn-glycero-3-Phosphocholine 455789062.4 343.12334 C12H22O11 1-alpha-D-Galactosyl-myo-inositol (lactose) 410172949.4 162.11236 C7H16NO3 L-Carnitine 404013964.3 383.31543 C23H42O4 [FA methyl(4:2/18:2)] methyl 9-butylperoxy-10_12-octadecadienoate 233919160.6 355.28412 C21H38O4 Methyl acetyl ricinoleate 190471829.2 160.1331 C8H18NO2 Methacholine 165776723.1 148.06034 C5H9NO4 L-Glutamate 151091045.9 222.09705 C8H15NO6 N-Acetyl-D-glucosamine 148793168.2 667.22894 C24H42O21 Glycogen 146868998.2 369.35138 C27H44 [ST] (5Z_7E)-9_10-seco-5_7_10(19)-cholestatriene 137976616.2 551.50347 C35H66O4 1-O-(1Z-Tetradecenyl)-2-(9Z-octadecenoyl)-sn-glycerol 121462416.9 313.27353 C19H36O3 [FA oxo(19:0)] 10-oxo-nonadecanoic acid 112684801.4 327.25284 C19H34O4 [FA methoxy_hydroxy(18:2)] 8-methoxy-13-hydroxy-9_11-octadecadienoic acid 108737293.9 218.13858 C10H19NO4 O-Propanoylcarnitine 93016510.12 299.22155 C17H30O4 [FA (17:1/2:0)] 8E-Heptadecenedioic acid 87738450.94 79.02126 C2H6OS Mercaptoethanol 75530497.29 177.05456 C10H8O3 1_3_8-Naphthalenertriol 69725634.82 232.15421 C11H21NO4 O-Butanoylcarnitine 66011561.24 Table 5-6: The top 20 metabolites by m/z intensity of positive ionisation mode analysis in control group. The mass-to-charge ratio, formula and the m/z intensity of each named metabolites are presented as well. Chapter 5, 227

E. coli row m/z Molecular formula Name m/z intensity 104.1071682 C5H14NO Choline 688544094.3 162.1123612 C7H16NO3 L-Carnitine 328583661.3 118.0865562 C5H11NO2 L-Valine 323402732.5 343.1233364 C12H22O11 1-alpha-D-Galactosyl-myo-inositol (lactose) 302781524.5 204.1228806 C9H17NO4 L-Acetylcarnitine 301572172.6 166.0861809 C9H11NO2 L-Phenylalanine 220244923.1 369.3513767 C27H44 [ST] (5Z_7E)-9_10-seco-5_7_10(19)-cholestatriene 208831576.3 86.09644175 C5H11N Piperidine 191708498.1 383.315431 C23H42O4 [FA methyl(4:2/18:2)] methyl 9-butylperoxy-10_12-octadecadienoate 179797044.2 355.2841233 C21H38O4 Methyl acetyl ricinoleate 156590246.4 160.1331034 C8H18NO2 Methacholine 144540343 667.2289361 C24H42O21 Glycogen 125929094.7 182.08111 C9H11NO3 L-Tyrosine 119735739.5 295.16499 C15H22N2O4 Leu-Tyr 113966956.8 116.070917 C5H9NO2 L-Proline 113953650.9 313.2735285 C19H36O3 [FA oxo(19:0)] 10-oxo-nonadecanoic acid 106969595.2 148.0603401 C5H9NO4 L-Glutamate 99580277.45 327.2528422 C19H34O4 [FA methoxy_hydroxy(18:2)] 8-methoxy-13-hydroxy-9_11-octadecadienoic acid 97136487.65 258.1099366 C8H21NO6P sn-glycero-3-Phosphocholine 86502311.46 299.2215456 C17H30O4 [FA (17:1/2:0)] 8E-Heptadecenedioic acid 77637189.42 Table 5-7: The top 20 metabolites by m/z intensity of positive ionisation mode analysis in E. coli group. The mass-to-charge ratio, formula and the m/z intensity of each named metabolites are presented as well. Chapter 5, 228

S. aureus row m/z Molecular formula Name m/z intensity 118.09 C5H11NO2 L-Valine 474193023 104.11 C5H14NO Choline 459760192 166.09 C9H11NO2 L-Phenylalanine 420899486 204.12 C9H17NO4 L-Acetylcarnitine 353171406 86.10 C5H11N Piperidine 317818082 245.19 C12H24N2O3 Leucyl-leucine 304379656 162.11 C7H16NO3 L-Carnitine 275815803 369.35 C27H44 [ST] (5Z_7E)-9_10-seco-5_7_10(19)-cholestatriene 266427751 116.07 C5H9NO2 L-Proline 253743805 343.12 C12H22O11 1-alpha-D-Galactosyl-myo-inositol 248665608 148.06 C5H9NO4 L-Glutamate 244795489 258.11 C8H21NO6P sn-glycero-3-Phosphocholine 231299639 182.08 C9H11NO3 L-Tyrosine 214276582 313.27 C19H36O3 [FA oxo(19:0)] 10-oxo-nonadecanoic acid 161755798 279.17 C15H22N2O3 Leu-Phe 154149870 295.16 C15H22N2O4 Leu-Tyr 153060034 318.20 C14H27N3O5 Leu-Val-Ser 151748149 344.25 C17H33N3O4 Leu-Leu-Val 139384457 233.15 C10H20N2O4 Leu-Thr 132935813 160.13 C8H18NO2 Methacholine 131531174 Table 5-8: The top 20 metabolites by m/z intensity of positive ionisation mode analysis in S. aureus group. The mass-to-charge ratio, formula and the m/z intensity of each named metabolites are presented as well. Chapter 5, 229

Control row m/z Molecular formula Name m/z intensity 191.02 C6H8O7 5-Dehydro-4-deoxy-D-glucarate (citrate) 1750330040 341.11 C12H22O11 1-alpha-D-Galactosyl-myo-inositol (lactose) 1513754934 300.05 C8H16NO9P N-Acetyl-D-glucosamine 6-phosphate 261652267 173.01 C6H6O6 cis-Aconitate 166893330 179.06 C6H12O6 myo-Inositol 106110746 178.05 C9H9NO3 Hippurate 98680533 167.02 C5H4N4O3 Urate 89863389 129.02 C5H6O4 2_5-Dioxopentanoate 84341727 266.09 C9H17NO8 Miserotoxin 78910119 215.03 C5H13O7P 2-C-Methyl-D-erythritol 4-phosphate 78482788 119.03 C4H8O4 L-Erythrulose 75351299 220.08 C8H15NO6 N-Acetyl-D-glucosamine 74489145 130.06 C4H9N3O2 Creatine 74407132 133.01 C4H6O5 (S)-Malate 73918031 146.05 C5H9NO4 L-Glutamate 73661846 333.06 C9H19O11P sn-glycero-3-Phospho-1-inositol 66333040 259.02 C6H13O9P D-Glucose 6-phosphate 52176193 147.03 C5H8O5 2-Dehydro-3-deoxy-L-arabinonate 44090003 85.03 C4H6O2 2-Butyne-1_4-diol 43558519 218.10 C9H17NO5 Pantothenate 43488200 Table 5-9: The top 20 metabolites by m/z intensity of negative ionisation mode analysis in control group. The mass-to-charge ratio, formula and the m/z intensity of each named metabolites are presented as well. Chapter 5, 230

E. coli row m/z Molecular formula Name m/z intensity 191.02 C6H8O7 5-Dehydro-4-deoxy-D-glucarate (citrate) 1375581553 341.11 C12H22O11 1-alpha-D-Galactosyl-myo-inositol (lactose) 1163878111 89.02 C3H6O3 (R)-Lactate 812425519 179.06 C6H12O6 myo-Inositol 182735001 117.02 C4H6O4 Succinate 179764837 173.01 C6H6O6 cis-Aconitate 105428739 215.03 C5H13O7P 2-C-Methyl-D-erythritol 4-phosphate 91564904 147.03 C5H8O5 2-Dehydro-3-deoxy-L-arabinonate 86395991 87.01 C3H4O3 Pyruvate 81948401 130.06 C4H9N3O2 Creatine 73380634 178.05 C9H9NO3 Hippurate 72131854 133.01 C4H6O5 (S)-Malate 71188741 215.03 C5H13O7P 2-C-Methyl-D-erythritol 4-phosphate 63327089 167.02 C5H4N4O3 Urate 60860379 146.05 C5H9NO4 L-Glutamate 60367636 129.02 C5H6O4 2_5-Dioxopentanoate 59941285 300.05 C8H16NO9P N-Acetyl-D-glucosamine 6-phosphate 59306858 171.01 C3H9O6P Glycerol 2-phosphate 51930002 145.01 C5H6O5 2-Oxoglutarate 47839083 175.06 C7H12O5 (2S)-2-Isopropylmalate 46199319 Table 5-10: The top 20 metabolites by m/z intensity of negative ionisation mode analysis in E. coli group. The mass-to-charge ratio, formula and the m/z intensity of each named metabolites are presented as well. Chapter 5, 231

S. aureus row m/z Molecular formula Name m/z intensity 191.02 C6H8O7 5-Dehydro-4-deoxy-D-glucarate (citrate) 1713079641 341.11 C12H22O11 1-alpha-D-Galactosyl-myo-inositol (lactose) 1246746397 89.02 C3H6O3 (R)-Lactate 921868860 146.05 C5H9NO4 L-Glutamate 155669416 173.01 C6H6O6 cis-Aconitate 124042799 179.06 C6H12O6 myo-Inositol (glucose/galactose) 116014627 133.01 C4H6O5 (S)-Malate 100377602 167.02 C5H4N4O3 Urate 94020844 215.03 C5H13O7P 2-C-Methyl-D-erythritol 4-phosphate 85004997 119.03 C4H8O4 L-Erythrulose 74942006 220.08 C8H15NO6 N-Acetyl-D-glucosamine 73840064 266.09 C9H17NO8 Miserotoxin 70788430 129.02 C5H6O4 2_5-Dioxopentanoate 69476176 124.01 C2H7NO3S Taurine 62376397 103.04 C4H8O3 (R)-3-Hydroxybutanoate 62310337 87.01 C3H4O3 Pyruvate 58442163 130.06 C4H9N3O2 Creatine 56614851 178.05 C9H9NO3 Hippurate 53708833 147.03 C5H8O5 2-Dehydro-3-deoxy-L-arabinonate 45261272 129.02 C5H6O4 Mesaconate 42718964 Table 5-11: The top 20 metabolites by m/z intensity of negative ionisation mode analysis in S. aureus group. The mass-to-charge ratio, formula and the m/z intensity of each named metabolites are presented as well. Chapter 5, 232

row Molecular formula Name m/z intensity (E. coli; EC) m/z intensity (S. aureus; SA) Difference in m/z intensity, m/z SA - EC 318.20 C14H27N3O5 Leu-Val-Ser 7169464 1.52E+08 144578685 392.25 C21H33N3O4 Leu-Leu-Phe 2515770 62511479 59995709 347.19 C14H26N4O6 Leu-Gln-Ser 66451.19 37865683 37799232 378.20 C19H27N3O5 Val-Pro-Tyr 77005.78 20353927 20276921 373.24 C17H32N4O5 Leu-Leu-Gln 68299.4 15413856 15345557 321.14 C11H20N4O7 Gln-Ser-Ser 24894.88 15198151 15173256 253.12 C12H16N2O4 Ala-Tyr 7303190 22111654 14808464 333.21 C14H28N4O5 Lys-Val-Ser 262397 13470687 13208290 496.22 C25H29N5O6 Trp-Gln-Tyr 381281.3 4664604 4283322 300.19 C14H25N3O4 Leu-Ala-Pro 139108.5 3942298 3803190 332.18 C14H25N3O6 Val-Val-Asp 1028657 4747132 3718476 318.17 C13H23N3O6 Leu-Ala-Asp 6637.732 3344439 3337801 423.22 C20H30N4O6 Leu-Gln-Tyr 338654.2 3321623 2982969 361.24 C16H32N4O5 Leu-Lys-Thr 5150.919 2567885 2562734 294.14 C14H19N3O4 Phe-Ala-Gly 1376626 3563815 2187189 409.21 C19H28N4O6 Leu-Asn-Tyr 3618.938 2079692 2076073 333.16 C16H20N4O4 Trp-Ala-Gly 115248 1737862 1622614 303.18 C11H22N6O4 Ala-Gly-Arg 213576.7 1810903 1597326 287.24 C14H30N4O2 N1_N12-Diacetylspermine 3925092 5423238 1498145 290.17 C12H23N3O5 Leu-Ala-Ser 17054.07 944656.8 927603 Table 5-12: Metabolites identified from positive ionisation mode analysis which were absent in control group but found greater in S. aureus infected group by means of m/z intensity than E. coli infected group. The m/z intensity for each of metabolites identified in both infected groups and the difference in m/z intensity were as presented. The difference in m/z intensity was equal to the m/z intensity of S. aureus infected group minus the m/z intensity of E. coli infected group Chapter 5, 233

row m/z Molecular formula Name m/z intensity m/z intensity Difference in (E. coli; EC) (S. aureus; m/z intensity, SA) SA - EC 187.12 C12H14N2 Calligonine 7135922 2485802 4650121 376.22 C20H29N3O4 Leu-Phe-Pro 15054435 11700662 3353774 181.07 C7H8N4O2 Theophylline 4655026 1465804 3189222 306.17 C12H23N3O6 Leu-Ser-Ser 1107760 283639 824122 407.22 C18H34N2O6S Lincomycin 1197998 434791 763207 199.09 C12H10N2O Harmalol 591018 108776 482242 408.25 C21H33N3O5 Leu-leu-tyr 1228049 841475 386574 350.21 C18H27N3O4 Leu-Phe-Ala 490555 130928 359627 468.34 C23H50NO6P [PC methyl(14:2)] 2-methyl-3-tetradecyl-sn-glycero-1- 731979 412006 319973 phosphocholine 295.18 C19H22N2O Eburnamonine 597554 286885 310669 334.16 C13H23N3O7 Leu-Asp-Ser 380049 88191 291859 286.18 C13H23N3O4 Leu-Gly-Pro 324809 36711 288099 165.07 C8H8N2O2 Ricinine 243149 4239 238910 458.25 C22H31N7O4 Trp-Pro-Arg 641744 451644 190100 140.03 C6H5NO3 4-Nitrophenol 171175 5689 165486 366.17 C17H23N3O6 Pro-Ser-Tyr 163275 14072 149204 296.12 C13H17N3O5 Asn-Tyr 655710 512128 143582 451.16 C20H26N4O6S Met-Trp-Asp 126010 1722 124289 393.14 C14H24N4O7S Met-Asp-Gln 164428 40184 124244 407.22 C18H34N2O6S Lincomycin 201509 80887 120622 Table 5-13: Metabolites identified from positive ionisation mode analysis which were absent in control group but found greater in E. coli infected group by means of m/z intensity than S. aureus infected group. The m/z intensity for each of metabolites identified in both infected groups Chapter 5, 234

and the difference in m/z intensity were as presented. The difference in m/z intensity was equal to the m/z intensity of S. aureus infected group minus the m/z intensity of E. coli infected group. Chapter 5, 235

row m/z Molecular formula Name m/z intensity (E. coli; EC) m/z intensity (S. aureus; SA) Difference in m/z intensity, SA - EC 345.18 C14H26N4O6 Leu-Gln-Ser 46782 18858881 18812099 331.20 C14H28N4O5 Lys-Val-Ser 640021 18012351 17372329 231.14 C10H20N2O4 Leu-Thr 189543 14873123 14683580 390.24 C21H33N3O4 Leu-Leu-Phe 258374 6241107 5982733 359.23 C16H32N4O5 Leu-Lys-Thr 34118 5490747 5456629 244.13 C10H19N3O4 Leu-Asn 145972 5592910 5446938 354.18 C15H25N5O5 Thr-Val-His 337 4768081 4767744 342.17 C15H25N3O6 Leu-Asp-Pro 50023 4469062 4419038 201.12 C9H18N2O3 Leu-Ala 303533 3822614 3519082 247.09 C9H16N2O6 Glu-Thr 618714 3901821 3283106 329.18 C14H26N4O5 Leu-Ala-Gln 11706 3261910 3250204 251.10 C12H16N2O4 Ala-Tyr 1199633 4388952 3189320 373.17 C15H26N4O7 Leu-Asp-Gln 27874 2290843 2262968 217.12 C9H18N2O4 Leu-Ser 1781005 3651567 1870562 311.14 C18H20N2O3 Phe-Phe 1020478 2422191 1401713 364.19 C18H27N3O5 Leu-Ala-Tyr 105095 1312711 1207616 345.21 C15H30N4O5 Leu-Lys-Ser 207570 1389000 1181430 205.08 C7H14N2O5 Thr-Ser 14626 1171204 1156578 350.17 C17H25N3O5 Leu-Gly-Tyr 147416 1294531 1147114 301.16 C11H22N6O4 Ala-Gly-Arg 204285 1220662 1016377 Table 5-14: Metabolites identified from negative ionisation mode analysis which were absent in control group but found greater in S. aureus infected group by means of m/z intensity than E. coli infected group. The m/z intensity for each of metabolites identified in both infected groups and the difference in m/z intensity were as presented. The difference in m/z intensity was equal to the m/z intensity of S. aureus infected group minus the m/z intensity of E. coli infected group. Chapter 5, 236

row Molecular Name m/z intensity (E. coli; m/z intensity (S. aureus; Difference in m/z intensity, SA - m/z formula EC) SA) EC 270.15 C12H21N3O4 Val-Gly-Pro 1214191 445816 768374 304.15 C12H23N3O6 Leu-Ser-Ser 399260 61590 337670 197.07 C7H10N4O3 5-Acetylamino-6-amino-3- 340796 54974 285821 methyluracil 277.12 C14H18N2O4 Pro-Tyr 520743 281163 239580 374.21 C20H29N3O4 Leu-Phe-Pro 1167418 987379 180040 147.01 C5H8O3S 4-Methylthio-2-oxobutanoic acid 154641 1714 152927 294.11 C13H17N3O5 Asn-Tyr 469728 328468 141260 355.20 C16H28N4O5 Leu-Gln-Pro 160771 42431 118340 281.11 C13H18N2O5 Thr-Tyr 408989 315714 93275 373.21 C16H30N4O6 Leu-Lys-Asp 903396 851149 52247 91.06 C7H8 Toluene 52963 996 51967 221.09 C11H14N2O3 Phe-Gly 95534 44609 50925 344.18 C15H27N3O6 Leu-Val-Asp 47648 16611 31037 217.12 C9H18N2O4 Leu-Ser 33732 11357 22375 395.20 C17H28N6O5 Leu-Gln-His 19559 3891 15667 284.16 C13H23N3O4 Leu-Gly-Pro 28598 17983 10615 Table 5-15: Metabolites identified from negative ionisation mode analysis which were absent in control group but found greater in E. coli infected group by means of m/z intensity than S. aureus infected group. The m/z intensity for each of metabolites identified in both infected groups and the difference in m/z intensity were as presented. The difference in m/z intensity was equal to the m/z intensity of S. aureus infected group minus the m/z intensity of E. coli infected group. Chapter 5, 237

5.3.4 Relative changes In order to identify the metabolites in both modes that showed differences between bacterial infection groups and control group, the median of the m/z intensities were calculated for each metabolite in each group. Fifteen metabolites that were either increased or decreased in both bacterial infection groups from control group in both modes are presented in Table 5-16, 5-17, 5-18 and 5-19. Most of the metabolites that increased in both bacterial infection groups were tripeptides or dipeptides (Val-Pro-Pro, Leu-Leu-Ser, Lys-Ala-Pro, Leu-Ala-Gln, Val-Arg, Leu-Val, etc).

The relative change in intensity for each identified metabolite between the sample groups from the infections was also calculated by dividing the median m/z intensity from the S. aureus samples by the median m/z intensity determined for the E. coli group. Fifteen metabolites that either increased or decreased when comparing between bacterial infection groups (Table 5-20 and 5-21). As the majority of the group data was non-parametric, the Mann-Witney test used to test for the significance of the difference between the bacterial groups. The tables (5-20 and 5-21) are arranged in increasing value of P as determined by the Mann-Witney test which shows the change observed in m/z intensity for the metabolites with the most significant difference in each of the categories. (Positive mode: significantly increased in S. aureus or increased in E. coli; Negative mode; significantly increased in S. aureus or in E. coli.) In positive mode, the p-values increased from 0 to 0.0439 in group of metabolites increased in E. coli and the p-values increased from 0.0001 to 0.037 in group of metabolites increased in S. aureus. In negative mode, the p-values increased from 0 to 0.0419 in group of metabolites increased in E. coli and the p-values increased from 0 to 0.0489 in group of metabolites increased in S. aureus. Chapter 5, 238

Reduced in .E coli row m/z Molecular formula Name m/z intensity (Control) m/z intensity (E. coli) Fold increase/decrease 221.09 C11H12N2O3 5-Hydroxy-L-tryptophan 396487.6 0 0 261.06 C14H12O3S Suprofen 235737.3 0 0 345.18 C14H24N4O6 Thr-Gln-Pro 405720.5 0 0 223.10 C12H14O4 Dillapiole 14979735 23468.75 0.002 177.05 C10H8O3 1_3_8-Naphthalenertriol 69725635 211373.6 0.003 419.32 C26H42O4 Diisononyl phthalate 14710838 45524.96 0.003 75.09 C3H10N2 1_3-Diaminopropane 469562.7 2367.125 0.005 79.02 C2H6OS Mercaptoethanol 75530497 1880468 0.025 160.10 C7H13NO3 (-)-Betonicine 1187418 31450.85 0.026 126.13 C8H15N gamma-Coniceine 284401.3 9156.271 0.032 208.10 C11H13NO3 Cantleyine 246641.8 8702.787 0.035 85.10 C6H12 Cyclohexane 1094320 42105.6 0.038 165.09 C10H12O2 Isoeugenol 661742.4 39724.09 0.060 268.10 C10H13N5O4 Deoxyguanosine 7124045 462562 0.065 142.03 C2H8NO4P Ethanolamine phosphate 505132.3 33286.07 0.066

Increased in E. coli 329.15 C18H20N2O4 Phe-Tyr 71 3960112 55932 283.13 C13H18N2O5 Thr-Tyr 48 1374349 28406 312.19 C15H25N3O4 Val-Pro-Pro 235 3696719 15739 188.18 C9H21N3O N1-Acetylspermidine 171 2211589 12931 313.15 C18H20N2O3 Phe-Phe 744 8893432 11959 Chapter 5, 239

274.19 C11H23N5O3 Val-Arg 855 9403370 10995 332.22 C15H29N3O5 Leu-Leu-Ser 78 753642 9678 315.20 C14H26N4O4 Lys-Ala-Pro 35 304693 8789 249.13 C10H20N2O3S Met-Val 103 867833 8393 217.12 C9H16N2O4 Thr-Pro 91 651287 7184 264.10 C9H17N3O4S Ala-Ala-Cys 115 492733 4274 366.20 C18H27N3O5 Leu-Ala-Tyr 227 801312 3524 272.16 C12H21N3O4 Val-Gly-Pro 2911 8509340 2923 175.11 C7H14N2O3 N-Acetylornithine 2353 5549482 2359 262.14 C10H19N3O5 Lys-Asp 25 55845 2267 Table 5-16: A relative change of metabolites in positive ionisation mode analysis in E. coli infected group in comparison to control group. The relative change in intensity for each identified metabolite between the sample groups was calculated by dividing the mean m/z intensity from the E. coli infected samples by the mean m/z intensity determined for the control group. Chapter 5, 240

Reduced in S. aureus row m/z Molecular formula Name m/z intensity (Control) m/z intensity (S.aureus) Fold increase/decrease 221.09 C11H12N2O3 5-Hydroxy-L-tryptophan 396487.6 0 n/a 261.06 C14H12O3S Suprofen 235737.3 0 n/a 232.17 C15H21NO Metazocine 230323.2 0 n/a 142.03 C2H8NO4P Ethanolamine phosphate 505132.3 0 n/a 212.04 C4H10N3O5P Phosphocreatine 20806449 0 n/a 152.06 C5H5N5O 8-Hydroxyadenine 7285295 0 n/a 183.09 C6H14O6 Mannitol 257290.9 0 n/a 174.02 C6H7NO3S 2-Aminobenzenesulfonate 6894.951 0 n/a 240.11 C9H13N5O3 Dihydrobiopterin 298827 0 n/a 223.10 C12H14O4 Dillapiole 14979735 25325.15 0.002 268.10 C10H13N5O4 Deoxyguanosine 7124045 17460.44 0.002 75.09 C3H10N2 1_3-Diaminopropane 469562.7 1188.912 0.003 131.13 C5H14N4 Agmatine 238399.8 813.1191 0.003 160.10 C7H13NO3 (-)-Betonicine 1187418 4261.909 0.004 177.05 C10H8O3 1_3_8-Naphthalenertriol 69725635 273655.9 0.004

Increased in S .aureus 329.15 C18H20N2O4 Phe-Tyr 71 6899649 97449 344.18 C15H25N3O6 Leu-Asp-Pro 370 24766005 66964 366.20 C18H27N3O5 Leu-Ala-Tyr 227 8501998 37387 Chapter 5, 241

331.20 C14H26N4O5 Leu-Ala-Gln 599 22347998 37289 274.19 C11H23N5O3 Val-Arg 855 30879290 36106 313.15 C18H20N2O3 Phe-Phe 744 25542108 34345 332.22 C15H29N3O5 Leu-Leu-Ser 78 2561617 32895 246.14 C10H19N3O4 Leu-Asn 290 8706973 30007 249.13 C10H20N2O3S Met-Val 103 2856985 27630 292.13 C14H17N3O4 Trp-Ser 188 4827665 25732 312.19 C15H25N3O4 Val-Pro-Pro 235 5247047 22340 302.21 C14H27N3O4 Leu-Leu-Gly 92 2001246 21736 283.13 C13H18N2O5 Thr-Tyr 48 1023145 21147 401.29 C18H36N6O4 Leu-Leu-Arg 213 3979172 18697 288.20 C12H25N5O3 Leu-Arg 1262 21946148 17391 Table 5-17: A relative change of metabolites in positive ionisation mode analysis in S. aureus infected group in comparison to control group. The relative change in intensity for each identified metabolite between the sample groups was calculated by dividing the mean m/z intensity from the S. aureus infected samples by the mean m/z intensity determined for the control group. Chapter 5, 242

Reduced in E. coli row m/z Molecular formula Name m/z intensity m/z intensity Fold increase/decrease (Control) (E. coli) 219.01 C7H8O8 4-Carboxy-4-hydroxy-2-oxoadipate 528160 2379 0.005 141.06 C7H10O3 4-Oxocyclohexanecarboxylate 129673 1436 0.011 112.99 C4H2O4 Acetylenedicarboxylate 143871 2380 0.017 275.02 C13H9N2O3Cl GW 9662 324140 12136 0.037 95.01 C5H4O2 Protoanemonin 59718 2432 0.041 155.03 C7H8O4 2-Hydroxy-6-keto-2_4-heptadienoate 231762 10357 0.045 210.03 C4H10N3O5P Phosphocreatine 1659260 89840 0.054 373.10 C18H18N2O7 PortulacaxanthinII 27781 1508 0.054 188.99 C6H6O5S 3-Sulfocatechol 88004 5368 0.061 113.02 C5H6O3 2-Hydroxy-2_4-pentadienoate 11126302 756437 0.068 153.02 C7H6O4 3_4-Dihydroxybenzoate 965175 99451 0.103 194.05 C9H9NO4 Salicyluric acid 1409008 163048 0.116 104.04 C3H7NO3 L-Serine 161750 19405 0.120 266.12 C17H17NO2 (-)-Caaverine 87469 10795 0.123 140.01 C2H8NO4P Ethanolamine phosphate 1079475 135707 0.126

Increased in E. coli 187.11 C8H16N2O3 N6-Acetyl-L-lysine 45 2013370 44736 259.13 C11H20N2O5 Glu-Leu 398 2881723 7249 327.14 C18H20N2O4 Phe-Tyr 37 263896 7187 195.05 C7H8N4O3 1-3-Dimethyluricacid 92 634581 6865 272.17 C11H23N5O3 Val-Arg 857 5143175 6004 339.09 C12H20O11 3-Ketosucrose 92 343018 3746 Chapter 5, 243

277.16 C15H22N2O3 Leu-Phe 132 392424 2962 466.33 C23H50NO6P [PC methyl(14:2)] 2-methyl-3-tetradecyl- 255 730878 2861 sn-glycero-1-phosphocholine 229.16 C11H22N2O3 Leu-Val 341 743738 2183 313.19 C14H26N4O4 Lys-Ala-Pro 151 310677 2060 181.04 C6H6N4O3 1-Methyluric acid 1359 2723161 2004 203.08 C11H12N2O2 L-Tryptophan 187 320534 1714 261.12 C14H18N2O3 Phe-Pro 525 855177 1628 243.10 C10H16N2O5 Glu-Pro 364 556272 1527 175.06 C7H12O5 (2S)-2-Isopropylmalate 31248 46199319 1478 Table 5-18: A relative change of metabolites in negative ionisation mode analysis in E. coli infected group in comparison to control group. The relative change in intensity for each identified metabolite between the sample groups was calculated by dividing the mean m/z intensity from the E. coli infected samples by the mean m/z intensity determined for the control group. Chapter 5, 244

Reduced in S. aureus row m/z Molecular Name m/z intensity m/z intensity Fold increase/decrease formula (Control) (S.aureus) 210.03 C4H10N3O5P Phosphocreatine 1659260 0 n/a 150.04 C5H5N5O 8-Hydroxyadenine 216939 0 n/a 219.01 C7H8O8 4-Carboxy-4-hydroxy-2-oxoadipate 528160 1524 0.003 140.01 C2H8NO4P Ethanolamine phosphate 1079475 7703 0.007 179.03 C9H8O4 3-(4-Hydroxyphenyl)pyruvate 53543 477 0.009 229.01 C5H11O8P D-Ribose 5-phosphate 655172 10906 0.017 155.03 C7H8O4 2-Hydroxy-6-keto-2_4-heptadienoate 231762 6764 0.029 259.02 C6H13O9P D-Glucose 6-phosphate 52176193 1588423 0.030 171.01 C3H9O6P Glycerol 2-phosphate 15741286 539478 0.034 141.06 C7H10O3 4-Oxocyclohexanecarboxylate 129673 5551 0.043 112.99 C4H2O4 Acetylenedicarboxylate 143871 6429 0.045 104.04 C3H7NO3 L-Serine 161750 9620 0.059 289.03 C7H15O10P D-Sedoheptulose 7-phosphate 1368021 81751 0.060 175.02 C6H8O6 D-Glucuronolactone 5130855 368321 0.072 298.07 C9H18NO8P D-4'-Phosphopantothenate 1604867 128723 0.080

Increased in S. aureus 187.11 C8H16N2O3 N6-Acetyl-L-lysine 45 8264384 183632 277.16 C15H22N2O3 Leu-Phe 132 13157998 99319 Chapter 5, 245

259.13 C11H20N2O5 Glu-Leu 398 9024354 22699 327.14 C18H20N2O4 Phe-Tyr 37 594526 16192 286.19 C12H25N5O3 Leu-Arg 304 4349308 14315 272.17 C11H23N5O3 Val-Arg 857 10020592 11697 243.17 C12H24N2O3 Leucyl-leucine 2334 23052455 9879 229.16 C11H22N2O3 Leu-Val 341 2683365 7875 233.08 C8H14N2O6 Glu-Ser 233 1236984 5305 203.08 C11H12N2O2 L-Tryptophan 187 786619 4207 241.09 C9H14N4O4 Ser-His 311 1043698 3360 321.24 C20H34O3 2alpha-(Hydroxymethyl)-5alpha-androstane- 496 1658760 3343 3beta_17beta-diol 345.24 C22H34O3 Taxa-4(20)_11(12)-dien-5alpha-acetoxy-10beta-ol 561 1687503 3010 232.13 C9H19N3O4 Lys-Ser 305 800527 2627 466.33 C23H50NO6P [PC methyl(14:2)] 2-methyl-3-tetradecyl-sn- 255 669343 2620 glycero-1-phosphocholine Table 5-19: A relative change of metabolites in negative ionisation mode analysis in S. aureus infected group in comparison to control group. The relative change in intensity for each identified metabolite between the sample groups was calculated by dividing the mean m/z intensity from the S. aureus infected samples by the mean m/z intensity determined for the control group.

Chapter 5, 246

Row m/z Molecular formula Name Control E. coli (EC) S. aureus (SA) SA/EC P value for EC.>SA E. coli > S. aureus 268.10 C10H13N5O4 Deoxyguanosine 7393670 246102 2422 0.010 0 360.17 C17H21N5O4 Phe-Gly-His 74624 147797 19243 0.130 0 131.13 C5H14N4 Agmatine 7274 25157 678 0.027 0.0001 262.13 C11H19NO6 Mycosporine 2215241 1490476 73639 0.049 0.0002 662.48 C35H68NO8P [PC (11:0/16:0)] 1-(10E-undecenoyl)-2-hexadecanoyl-sn-glycero-3-phosphocholine 807946 1369890 82201 0.060 0.0004 395.32 C24H42O4 [ST (4:0)] 5alpha-Cholane-3alpha_7alpha_12alpha_24-tetrol 675090 237003 28238 0.119 0.0007 347.17 C14H26N4O4S Lys-Cys-Pro 0 14053 2309 0.164 0.0008 302.15 C16H19N3O3 Trp-Pro 3722 60648 6427 0.106 0.001 690.51 C37H72NO8P [PC (18:0/11:0)] 1-octadecanoyl-2-(10E-undecenoyl)-sn-glycero-3-phosphocholine 6104185 6834848 1086403 0.159 0.0012 122.10 C8H11N Phenethylamine 332351 366314 28392 0.078 0.0013 136.06 C5H5N5 Adenine 15149 25308 2037 0.080 0.0015 170.08 C8H11NO3 Pyridoxine 10690 43702 4123 0.094 0.0035 292.10 C11H17NO8 2_7-Anhydro-alpha-N-acetylneuraminic acid 392316 242233 30159 0.125 0.0061 341.27 C20H36O4 [FA (20:2)] 15S-hydroperoxy-11Z_13E-eicosadienoic acid 3284897 1757064 278653 0.159 0.0077 415.23 C22H30N4O4 Tentoxin 4205 873662 59028 0.068 0.0185 247.11 C10H18N2O3S Met-Pro 83543 3674935 453681 0.123 0.0307 291.12 C11H18N2O7 N-Succinyl-LL-2_6-diaminoheptanedioate 132015 45106 6421 0.142 0.0371 452.28 C21H42NO7P LysoPE(0 22523 429336 70584 0.164 0.0439

Chapter 5, 247

Row m/z Molecular formula Name Control E. coli (EC) S. aureus (SA) SA/EC P value for EC.>SA S. aureus > E. coli

288.20 C12H25N5O3 Leu-Arg 882 106405 2643318 24.8 0.0001 295.18 C19H22N2O Eburnamonine 0 7322 911037 124.4 0.0001 204.13 C8H17N3O3 Lys-Gly 0 4545 105894 23.3 0.0002 275.17 C11H22N4O4 Lys-Ala-Gly 0 7408 259091 35.0 0.0007 288.22 C15H29NO4 L-Octanoylcarnitine 32223 21165 882738 41.7 0.0011 217.13 C8H16N4O3 N-a-Acetyl-L-arginine 17622 6593 251098 38.1 0.0012 316.25 C17H33NO4 O-Decanoyl-L-carnitine 6766 11711 528729 45.1 0.0012 233.15 C10H20N2O4 Leu-Thr 1694 599153 13755161 23.0 0.0042 93.05 C3H8O3 Glycerol 948 2760 301366 109.2 0.0049 262.10 C9H15N3O6 Ala-Asp-Gly 0 2511 527607 210.1 0.0062 243.11 C9H14N4O4 Ser-His 0 3427 111584 32.6 0.0067 246.16 C9H19N5O3 beta-Alanyl-L-arginine 0 595 21148 35.5 0.0086 304.16 C11H21N5O5 Glu-Arg 0 1208 57459 47.6 0.0102 246.14 C10H19N3O4 Leu-Asn 0 6124 1226699 200.3 0.0139 220.09 C7H13N3O5 1D-1-Guanidino-1-deoxy-3-dehydro-scyllo-inositol 0 401 116297 290.2 0.0162 400.34 C23H45NO4 L-Palmitoylcarnitine 0 49632 1440263 29.0 0.0166 300.29 C18H37NO2 Sphingosine 10283 11327 975206 86.1 0.0185 Chapter 5, 248

428.37 C25H49NO4 Stearoylcarnitine 3663 31932 870431 27.3 0.0203 241.03 C6H12N2O4S2 L-Cystine 233951 15219 1219633 80.1 0.0228 320.12 C15H17N3O5 Trp-Asp 0 1689 186229 110.2 0.037

Table 5-20: A relative change of metabolites in positive ionisation mode analysis in E.coli and S. aureus infected groups in comparison to control group. The relative change in intensity for each identified metabolite between the sample groups from the infections was also calculated by dividing the median m/z intensity from the S. aureus samples by the median m/z intensity determined for the E. coli group.

Chapter 5, 249

Row m/z Molecular formula Name Control E. coli (EC) S. aureus (SA) P- value SA/EC E. coli > S. aureus 377.11 C13H22N4O7S Met-Asn-Asp 6890577.34 8802326.18 214344.33 0.024 0 421.22 C23H34O7 Picrasin C 14239.19 41027.42 3517.35 0.086 0 181.05 C9H10O4 3-Methoxy-4-hydroxyphenylglycolaldehyde 62009.99 529282.74 39401.15 0.074 0.0001 149.02 C8H6O3 2-Carboxybenzaldehyde 1776.99 19169.73 3885.44 0.203 0.0002 208.06 C7H15NO4S MOPS 463757.13 674283.55 55062.11 0.082 0.0003 413.22 C22H30N4O4 Tentoxin 0.00 9914.58 684.99 0.069 0.0006 227.14 C11H20N2O3 Leu-Pro 0.00 586049.23 103807.18 0.177 0.0012 175.06 C7H12O5 (2S)-2-Isopropylmalate 28034.31 484537.27 16910.05 0.035 0.0013 185.09 C8H14N2O3 Ala-Pro 0.00 110493.93 4949.62 0.045 0.0015 225.05 C9H10N2O5 3-Nitrotyrosine 16030.61 64844.81 8919.53 0.138 0.0015 191.06 C7H12O6 Valiolone 459049.45 836759.32 142111.11 0.170 0.0022 464.28 C22H44NO7P [PC (14:0)] 1-(9Z-tetradecenoyl)-sn-glycero-3-phosphocholine 6635.16 61018.85 9555.26 0.157 0.0026 240.05 C10H11NO6 N-(2_3-Dihydroxybenzoyl)-L-serine 4933.71 12066.26 1245.97 0.103 0.0031 241.01 C6H11O8P D-myo-Inositol 1_2-cyclic phosphate 126772.78 268187.98 18729.54 0.070 0.0037 138.02 C6H5NO3 4-Nitrophenol 2779.26 266216.76 18730.86 0.070 0.0068 157.05 C7H10O4 2-Isopropylmaleate 9008.42 120032.46 18651.52 0.155 0.0077 163.04 C9H8O3 Phenylpyruvate 4799.75 188677.58 36508.59 0.193 0.0077 304.15 C12H23N3O6 Leu-Ser-Ser 0.00 77330.97 9092.19 0.118 0.0108 166.04 C5H5N5O2 8-Hydroxyguanine 284694.89 163094.08 17370.75 0.107 0.0166 181.04 C6H6N4O3 1-Methyluric acid 1341.04 237648.75 1724.51 0.007 0.0302 171.01 C3H9O6P Glycerol 2-phosphate 12866412.16 3984791.07 350936.16 0.088 0.0339 91.06 C7H8 Toluene 0.00 3206.91 574.50 0.179 0.0419 Chapter 5, 250

Row m/z Molecular formula Name Control E. coli (EC) S. aureus (SA) P-value SA/EC S. aureus > E. coli 111.01 C5H4O3 3-Furoicacid 144280.02 37867.85 1270896.79 33.6 0 241.09 C9H14N4O4 Ser-His 0.00 5983.68 404950.12 67.7 0.0007 315.20 C14H28N4O4 Leu-Lys-Gly 0.00 2223.93 50059.68 22.5 0.0015 237.09 C11H14N2O4 Gly-Tyr 0.00 7716.76 145858.86 18.9 0.0024 339.25 C20H36O4 [FA (20:2)] 15S-hydroperoxy-11Z_13E-eicosadienoic acid 5925.74 7410.93 161970.79 21.9 0.0025 302.15 C11H21N5O5 Glu-Arg 0.00 1449.54 51160.10 35.3 0.0028 246.15 C10H21N3O4 Lys-Thr 203.59 72670.00 2556646.63 35.2 0.0042 260.09 C9H15N3O6 Ala-Asp-Gly 0.00 4508.99 1183200.27 262.4 0.0042 227.20 C14H28O2 Tetradecanoic acid 1803.83 7008.93 176840.89 25.2 0.0048 360.14 C14H23N3O8 Leu-Asp-Asp 0.00 1015.08 78304.48 77.1 0.0094 255.23 C16H32O2 Hexadecanoic acid 24943.22 57428.79 1014884.73 17.7 0.0108 203.07 C7H12N2O5 Ala-Asp 0.00 10488.46 522376.69 49.8 0.0145 239.02 C6H12N2O4S2 L-Cystine 85659.22 2122.87 560196.97 263.9 0.0161 260.13 C10H19N3O5 Lys-Asp 0.00 1026.23 56647.44 55.2 0.0185 301.24 C17H34O4 MG(0 3153.57 4636.33 93332.57 20.1 0.0278 278.05 C8H14N3O6P Cidofovir 655109.72 136903.56 2662709.08 19.4 0.0283 231.10 C9H16N2O5 N2-Succinyl-L-ornithine 0.00 1513.78 46034.74 30.4 0.0293 233.08 C8H14N2O6 Glu-Ser 0.00 7358.12 491377.73 66.8 0.0303 253.03 C7H14N2O4S2 L-Djenkolic acid 0.00 762.58 15942.83 20.9 0.0324 218.08 C7H13N3O5 1D-1-Guanidino-1-deoxy-3-dehydro-scyllo-inositol 0.00 738.01 101885.03 138.1 0.038 353.31 C22H42O3 [FA oxo(22:0)] 10-oxo-docosanoic acid 1592.50 6321.76 113698.29 18.0 0.0489 Chapter 5, 251

201.12 C9H18N2O3 Leu-Ala 0.00 32485.10 869131.79 26.8 0.0644 258.15 C11H21N3O4 Leu-Ala-Gly 0.00 38433.25 666891.94 17.4 0.083 205.08 C7H14N2O5 Thr-Ser 0.00 3997.60 199953.30 50.0 0.0963 300.05 C8H16NO9P N-Acetyl-D-glucosamine 6-phosphate 250003647.30 373727.48 19134245.55 51.2 0.2508

Table 5-21: A relative change of metabolites in negative ionisation mode analysis in E.coli and S. aureus infected group in comparison to control group. The relative change in intensity for each identified metabolite between the sample groups from the infections was also calculated by dividing the median m/z intensity from the S. aureus samples by the median m/z intensity determined for the E. coli group. Chapter 5, 252

5.3.5 Metabolic Pathways Altered by Bovine Mastitis

Pathos web facility provided by the Scottish Metabolomics Facility was used to further analyze the metabolites identified in each group of milk samples. The file was downloaded and this website (http://motif.gla.ac.uk/Pathos/index.html) allowed an investigation of the metabolites found in certain group of samples according to corresponded metabolic pathways. The pathways were listed starting from the greatest changes in the metabolites abundances in mastitic samples to the least changes in the metabolites abundances in both groups of mastitic as presented in Figure 5-1 and 5-2. The colour-coding scale indicator at each of the metabolites showed their abundances from increase to decrease in the milk samples. According to the website, light blue colour is an indication of decreased of metabolites in mastitic milk samples whilst the dark red colour is an indication of increased of metabolite in mastitic milk samples.

On each of the metabolic pathway, all metabolites involved in the particular metabolic pathway can be analyzed with the presented histogram showing the abundances in three different groups of milk samples as shown in Figure 5-3. A KEGG map was also be generated to highlight the potential metabolites involved in a particular metabolic pathway as showing in Figure 5-4 for the arachidonic acid pathway.

The next step involved was dividing the metabolic pathways identified into three different groups; lipid, protein and carbohydrate metabolisms. Six metabolic pathways with the most apparent changes in each metabolism from both positive and negative ionisation modes were chosen based on the abundances values provided by the Pathos website. For the lipid metabolism (Table 5-22), the arachidonic acid metabolic pathway was the pathway with most apparent changes with icosatetraenoic acid, prostaglandin F2 alpha, prostaglandin A2, leukotrienes A4 and B4 were increased in mastitic samples whilst no metabolites related to this pathway was found decreased in mastitis. Linoleic acid and α- linolenic acid metabolic pathways were the next pathways with the most changes with increased in icosatetraenoic and stearidonic acids in mastitic samples. Hexadecanoic acid was also found to increase in mastitic milk which derived from fatty acid-related metabolic pathways. The majority of these metabolites were the main precursors of eicosanoids involved in inflammatory activity in a host. Thus, it is not surprising to see these metabolites increased in mastitic samples. Chapter 5, 253

For carbohydrate metabolism (Table 5-23), galactose, pentose and glucuronate interconversions, fructose and mannose, starch and sucrose metabolic pathways were the pathways with the most apparent changes in carbohydrate metabolism apart from glycolysis/gluconeogenesis pathway. Glycerol, glycerone phosphate, fructose and glucose were the metabolites found increased in mastitic samples, whilst fructose-6-phosphate, mannose-6-phosphate and glucose-6-phosphate were among the metabolites decreased in mastitic samples.

Protein metabolism was analyzed and the metabolic pathways with the most apparent changes were arginine, tryptophan, tyrosine, phenylalanine, cysteine and methionine, as well as glycine, serine and threonine metabolic pathways (Table 5-24). The metabolites found to be increased in mastitic samples were tryptophan, ornithine, methionine and tyrosine. Glutamine, phosphocreatine, acetoacetate, propanoate and serine were among the protein metabolites decreased in mastitic milk samples.

Di- and tri-peptides were also found to increase in both groups of mastitic samples. A table of peptides identified by Pathos programme is presented in Table 5-25 and includes Asn- His, Asp-Arg, Gln-Ser-Ser and Leu- Gly- Ser. Chapter 5, 254

Figure 5-1: An example of various metabolic pathways generated by PATHOS website. In this example, the data from negative ionisation mode was used and the metabolic pathway was listed starting with the most apparent changes of metabolites to the least apparent changes of metabolites. Chapter 5, 255

Figure 5-3: An example of histogram of prostaglandin F2 alpha showing abundances in control, E. coli infected and S. aureus infected groups. It shows here that prostaglandin F2 alpha increased in abundances in both mastitis groups but present in the lowest abundances in control group.

Chapter 5, 257

Figure 5-4: An example of KEGG map generated from arachidonic acid metabolism to highlight the potential metabolites involved in a particular metabolic pathway. A dark red dot shows metabolite which was increased in mastitic milk samples. Further click on the dot revealed a histogram as showed in Figure 5.2. Chapter 5, 258

Lipid Metabolic Pathways Metabolites Increased in Mastitis Metabolites Decreased in Mastitis

Arachidonic acid Met. Icosatetraenoic acid, Prostaglandin F2α, None Leukotriene A4, Leukotriene B4, Prostagladin A2

Linoleic acid Met. Icosatetraenoic acid, Linoleate

α-linolenic acid Met. Stearidonic acid Heptadecatrienal

Biosynthesis of unsaturated fatty acid Hexadecanoic acid Icosatetraenoic acid

Fatty acid biosynthesis Hexadecanoic acid None

Fatty acid Met. Hexadecanoic acid None

Table 5-22: Six metabolic pathways in lipid metabolism with the most apparent changes of metabolites from both positive and negative ionisation mode analyses. The abundances values were provided by the Pathos website. Chapter 5, 259

Carbohydrate Metabolic Pathways Metabolites Increased in Mastitis Metabolites Decreased in Mastitis

Galactose Met. Glycerol D-Glucose-1-Phosphate, α-D-Galactose 1-phosphate

Pentose & Glucuronate Glycerone phosphate, D-Xylonolactone D-Fructose-6-Phosphate interconversions

Fructose & Mannose Met D-Fructose D-Mannose-6-Phosphate

Starch & Sucrose Met. Reduced acceptor D-Glucose D-Glucose-6-phosphate

Glycolysis/Gluconeogenesis None Acetate

Table 5-23: Six metabolic pathways in carbohydrate metabolism with the most apparent changes of metabolites from both positive and negative ionisation mode analyses. The abundances values were provided by the Pathos website. Chapter 5, 260

Protein Metabolic Pathways Metabolites Increased in Mastitis Metabolites Decreased in Mastitis

ArginineMet. N-Acetylornithine, L-Ornithine L-Glutamine, Phosphocreatine

Tryptophan Met. L-Tryptophan Anthranilate, 2-Oxoadipate

Tyrosine Met. L-Tyrosine, 4- Acetoacetate Hydroxyphenylacetate

Phenylalanine Met. L-Phenylalanine, Succinate Propanoate, 2-Hydroxy-2,4- pentadienoate

Cysteine and Methionine Met. L-Methionine 2-Oxobutanoate, L-Serine

Glycine, Serine and Threonine Met. L-Tryptophan L-Serine,D-serine

Table 5-24: Six metabolic pathways in protein metabolism with the most apparent changes of metabolites from both positive and negative ionisation mode analyses. The abundances values were provided by the Pathos website. Chapter 5, 261

Amino acid/Peptides Examples

Di-peptides Asn-His, Asp-Arg, Lys-Asn, Leu-Met, Ala-Tyr, Trp-Gly

Tri-peptides Gln-Ser-Ser, Leu-Gly-Ser, Met-Ala-Ser, Ala- Gly-Arg, Leu-Asp-Gly, Leu-Asn-Gly

Table 5-25: Examples of di- and tri-peptides which were found increased in both mastitic groups from both positive and negative ionisation mode analyses. The abundances values were provided by the Pathos website. Chapter 5, 262

5.4 Discussion From positive ion mode analysis, the 3 metabolites with highest intensity in the control group (Table 5-4) were L-acetylcarnitine, choline and sn-glycero-3-phosphocholine which are related to lipid metabolism and the choline content of phopsholipids in the milk fat globule membrane. Citrate, lactose, glucose and galactose were found high in intensity from negative ion mode in both control and E. coli infected group but the monosaccharides were not at the top in S. aureus infected group, possibly indicative that these monosaccharides are more rapidly absorbed for use in metabolism by the latter bacterium

Identification of the metabolites which show the most significant differences between the infections caused by the bacteria provide valuable indications for demonstration of differences in the metabolic effects of the infections and also reveal potential targets for biomarkers to identify the type of bacteria causing the mastitis. Notable in the evaluation of metabolite changes on infection are the increase in identification of di-and tripeptides in in both infections with particular tripeptides being associated with both of the bacteria, though with S. aureus infection causing more of these peptides to be present. The source of the peptides in the infected samples is likely to be breakdown products from milk proteins caused by bacterial or immune cell protease activity. The finding that there are differences between bacterial causative agent could be either due to differences in the specifity of action of the bacterial proteases or that the bacteria have differing efficiencies in absorbing the peptides after they are released by hydrolysis. As S. aureus infection caused the greater degree of di and tripeptide in the metabolite pool, the bacteria either have more active proteolytic enzymes or are less efficient at absorbing the peptides for their own metabolism.

An attempt was made to find the origin of the tripeptides by comparison to the bovine genome of those found in either infection group. Although 40% of the tripeptides present in the mastitis milk could be identified in the amino acid sequence of the milk proteins: caseins, α-lactalbumin, β-lactoglobin, albumin and lactoferrin it was found by scanning the bovine protein database, that the tri-peptides were identified in over 5000 proteins in the bovine proteome so that no definitive conclusion could be made on the origin of the peptides.

Of the 100 metabolites investigated, with the largest differences in m/z intensity between S. aureus and E. coli, there was a mix of metabolites, with lipid, nucleotide and carbohydrate related metabolites being identified. There were also non-host and non- Chapter 5, 263 pathogen metabolites identified such as tentoxin, ricinine, quinate, toluene and others. When these occurred in the infected groups and not in the control group it is possible that the metabolites are derived from therapeutic reagents. A number had metabolite database hits as either drugs (for infection or inflammation) or of compounds used in herbal or chinese medicine. As mentioned above the identification by the computer database of a number of metabolites were questionable, and while a number of these were identified and corrected (1-alpha-D-galactosyl-myo-inositol =lactose; myo-inositol = glucose or galactose) it is possible that the molecules are incorrectly named and these identified agents may be erroneously named. Confirmation of molecular identity would be a major task in following up this initial investigation.

5.4.1 Lipid Metabolism As part of the analysis, the effect of mastitis on lipid metabolism was assessed by examination of the metabolites in the 6 metabolic pathways with the most apparent changes from both positive and negative ion modes. From these six metabolic pathways, the 6 metabolites greater in both diseased and control groups were selected for discussion.

The method reported in this chapter revealed increases in arachidonic acid metabolites in both two mastitic group caused by E. coli and S. aureus. Examples of increased metabolites are icosatetraenoic acid, prostaglandin F2α (PGF2α), prostaglandin E2

(PGE2), leukotriene A4 (LKA4), leukotriene B4 (LKB4), thromboxane A2 (TXA2), thromboxane B2 (TXB2) and others.

These metabolites are the main precursors of eicosanoids which are the fatty acid-derived mediators involved in a wide-range of inflammatory activity in the host cell (Bannerman 2009). Thromboxanes and prostaglandins are formed by the catabolism of arachidonic acid mediated by cyclooxygenase while the formation of leukotrienes is mediated by lipoxygenase (Kuehl Jr. and Egan 1980; Williams and Higgs 1988). Each eicosanoids has a differing role in inflammatory processes and may have antagonizing effects to other eicosanoids. Thromboxane A2 for example is most potent of the eicosanoids in contracting aortic tissue and can trigger platelet aggregation whereas other prostaglandins (e.g, prostacyclin) and PGI2 cause vasodilation and inhibit platelet aggregation (Kuehl Jr. and Egan 1980; Ball et al. 1986; Feuerstein and Hallenbeck 1987). Meanwhile, leukotrienes are involved in inducing leukocyte recruitment and promoting leukocyte respiratory burst activity and cytokine production (Zipser and Laffi 1985; Williams and Higgs 1988; Bottoms and Adams 1992). According to Bannermann (2008), leukotriene B4 is the most Chapter 5, 264 potent endogenous chemoattractant. Thus, eicosanoids are responsible in mediating inflammatory response with a wide-range of activities on the immune response towards infection and injury.

With their important role as mediators in inflammatory responses, it is not surprising that increases in concentrations of these eicosanoids happen in bovine mastitis. These results were in accordance with Giri et al. (1984) in which they also found a greater concentration of both PGF1α and PGF2α in milk after 50μg infusion of LPS. However, they did not find any increase in other eicosanoids such as TXB2 or LKB4. Anderson et al. (1985) on the other hand, revealed only an increase of TXB2 milk level after 10μg of LPS infusion. According to Anderson et al. (1985), the reason of an absence of PGF2α in the milk samples following infusion in their study is because of the reduced concentration of LPS infused into the quarters as compared to the former study.

In this present study, both of the different pathogen-causing mastitis; E. coli and S. aureus are able to induce the release of these eicosanoids. One experimental intrammary infection study on E. coli by Peter et al. (1990) showed increases in prostacyclin (PCM), PGE2 and TXB2. However, no PCM was detected in this present study. However, Anderson et al (1985) also failed to show any increased in both PGF2α and TXB2 in experimental study of E.Coli. These differences are due to large biological and individual variability observed in milk eicosanoids concentrations in cows with acute coliform mastitis.

Furthermore, studies by Atroshi et al. (1986, 1987) which involved naturally occurring cases of mastitis caused by S. uberis, S. aureus, S. dysgalactiae or Micrococcus species revealed increases in PGE2, PGF2α and TXB2. In this present study, milk samples infected with S. aureus showed greater concentrations of both TXA2 and TXB2 than milk samples infected with E. coli. Similarly, study by Boutet et al. (2003) on chronic mastitis caused by coagulase-negative staphylococci such as S. uberis, S. agalactiae or S. aureus showed increases in LTB4. Based on these results, it is indicated that different types of pathogen elicit increases in releasing eicosanoids during bovine mastitis.

Apart from increased level of eicosanoids in all mastitic milk samples, there was also an increased in fatty acids metabolites seen in both groups of mastitic samples such as hexadecanoic acid, icosatetraenoic acid, linoleate, octadecenoic acid, octadecanoic acid, tetradecanoic acid etc.

Fatty acids in the blood play an important role in modulating inflammatory reactions (Calder 2008; Serhan et al. 2008). Stored fatty acids are important as energy substrate Chapter 5, 265 especially during negative energy balance (NEB) period. During this period, in a lactating cow the majority of available glucose is directed to the mammary gland for lactose synthesis (Herdt 2000). Thus, lipid mobilization from body reserve occurs and as a consequence, there is a release of non-esterified fatty acids (NEFA) into the blood stream. However an excess in the levels of circulating NEFAs could contribute to increase in disease susceptibility dairy cows especially in periparturient inflammatory-based diseases including mastitis, ketosis and metritis (Bernabucci et al. 2005; Goff 2006; Douglas et al. 2007).

Fatty acids can either provide benefit or have detrimental effects on the host through its regulating mechanisms in inflammatory reactions (Sordillo et al. 2009). For example, saturated fatty acids including lauric, palmitic and oleic acids can activate the toll-like receptor 4 (TLR4) through activation of NK-κB which leads to activation of innate immune response of the host. Thus, the presence of hexadecanoic acid or palmitic acid; a saturated fatty acid in both groups of mastitic samples could play an important role in activation of host innate immune response which provide the initial protection against invading pathogens before the acquired immune response is activated.

During inflammation of the udder, milk composition is known to change and this includes the fat content in the milk. Previous studies showed that the levels of free fatty acids (FFA) is increased during mastitis (Randolph & Erwin, 1979; Salih & Anderson 1979; Needs & Anderson 1984; Atroshi et al. 1989). These events were contributed to a high lipase activity which leads to lipolysis and increase in the permeability of the mammary gland to blood lipids (Randolph & Erwin 1979). According to Salih & Anderson (1979), lipoprotein lipase (LPL) is an enzyme that responsible for lipolysis in milk. Furthermore, the transfer of an increased amount of LPL activator from blood to milk may contribute to partly caused rise of the FFA in mastitic milk.

However, a study by Fitz-Gerald et al. (1981) did not agree on the role of LPL in increasing the level of FFA in mastitic milk. It might be explained by other mastitis- influenced factors (e.g; presence of blood-serum activators or changes in substrate susceptibility). Lower LPL activity in mastitic milk is due to impaired biosynthetic capability of mammary gland through permanent tissue damage or the action of proteolytic enzymes which are also present in the milk during mastitis. This was also supported by Anderson and Needs (1983) who observed high levels of FFA from lipolysis in mastitic milk but these increases were not necessarily due to LPL activators. Blood serum or heparin might give a stimulatory effect on lipolysis as discussed by Salih & Anderson Chapter 5, 266

(1979) previously. According to Murphy et al. (1989), increase in lipolysis was contributed from the susceptibility of the milk fat substrate to lipolysis in mastitic milk and not from the high level of lipase acitivity.

A number of factors can influence the level of FFA in milk such as above-normal FFA levels on secretion, LPL activator, cell lipase, proteolytic enzymes, anions, heparin-like substances and fat globule membrane (Salih & Anderson 1979; Fitz-Gerald et al. 1981). Other than that, the variation occurred in individual cows on the levels of FFA in milk are also influenced by factors such as diet, oestrus and stage of lactation (Anderson & Needs 1983).

In this present study, both linoleic and α-linolenic acid metabolism were increased in both mastitic groups. This is with an agreement with Hogan et al. (1987) in which they observed high content of unsaturated linoleic and linolenic acid in mastitic milk which might be important in the killing of bacteria. Linoleic acid is a precursor of archidonic acid and both could be released from the cell membrane phospholipids by phospholipase-like enzymes. Futhermore, according to Atroshi et al. (1989), α-linolenic acid was positively correlated with PGE2 and PGF2α which were also found in higher levels in mastitic milks in this present study than in healthy milk.

In this present study, high levels of long-chain fatty acids were observed in both groups of mastitic milk. However, it is not an agreement with studies by Kisza & Batura (1969), Randolph & Erwin (1974) and Karijord et al. (1982) only reported the highest content of 18:0 and C 18:1 in milk during early lactation but not during mastitis.

In contrast, other previous studies argued with the high fat content observed in mastitic milk (Bartsch et al. 1981; Massart-Leen et al. 1994). They assumed that during mastitis, the low level in milk yield could lead to reduce synthesis of the total amount amount of milk fat as the mammary gland epithelium does not retain its normal integrity for synthesis of major milk components.

5.4.2 Carbohydrate Metabolism Carbohydrates are one of the most important sources of energy in all living organisms. In cows, carbohydrate is fermented by microorganisms in the rumen to produce volatile fatty acids such as acetic acid, butyric acid and propionic acid. Gluconeogenesis in the cows occurs in the liver and glucose is produced from metabolites such as propionate, valerate, amino acids, lactate and glycerol (Nafikov & Beitz 2007). Chapter 5, 267

The mammary gland takes up glucose from the liver via circulatory system to be used in lactose synthesis. Glucose is also important for lipid metabolism as it serves as the backbone of milk fat synthesis in which glucose is converted to glycerol. Glycerol is then being attached to the fatty acids produced by acetate and ß-hydroxybutyrate. About 70% of net uptake of glucose from blood by lactating bovine udders is used to synthesize lactose and the remainder is oxidized in pentose and TCA cycles (Qiao et al. 2005).

Following mass spectrometry of metabolomics, six carbohydrate metabolic pathways were chosen and in each metabolic pathway, metabolites that increased or decreased in mastitis in relation to control animals were assigned in both positive and negative ion modes.

It was apparent that some carbohydrate metabolites were less evident in both groups of mastitis compared to control animals. Among these metabolites are glycerol, D-fructose, and others, while D-glucose or its isomer D-galactose and identified by the PATHOS programme as myo-inositol was still in the top 4 in E. coli and top 6 in S. aureus groups. Marschke & Kitchen (1984) agreed that levels of glucose were negatively correlated with both log somatic cell counts (logSCC) and log N-acetyl-beta-D-glucosaminidase (logNAGase). However, the reason for low levels of glucose found in mastitic milk is unknown. Several hypotheses have been made on how this event occurred. According to Linzell & Peaker (1971), during mastitis, reduced blood flow to the udder result in reduced in the supply of blood glucose to the secretory cells of mammary gland. It is also possible that glucose would move out of milk during mastitis by a paracellular pathway as Na+ and Cl- increased in milk. This would to help in maintaining the osmolarity between extracellular fluid and milk (Marschke & Kitchen 1984).

A study by Cecil et al. (1965) revealed that glycogen concentrations were greatly increased in mastitic milk compared with normal milk. As leukocytes contain high glycogen levels, there is a possibility that high glucose levels in mastitic milk are due to leukocyte infiltration during the inflammatory processes. However, the results in this present study showed that the end-products of glycogen breakdown such as glucose-1-phosphate and glucose-6-phosphate are lower in mastitic milk than in control milk. It seems that the influence of somatic cells in mastitis milk samples used in the present study was not great enough to release glycogen in the milk. Futhermore, because glycogen is only available from the liver as glucose for it to become accessible to other organs, reduced blood flow to the mammary gland during mastitis was the reason why glycogen levels are reduced in mastitic milk. Chapter 5, 268

Giesecke & Heever (1981) observed a fluctuation in glucose levels in milk from artificially-induced mastitis by human strains Group B Streptococci intracisternally. In certain conditions, low levels of glucose may occur relative to its supply and demand even in normal udder quarters. Reduced concentration of glucose in mastitic milk can affect the abilities of PMN-leukocytes by affecting the efficiency of phagocytosis and the killing of bacteria (Giesecke & Heever 1981).

Low levels of glucose contribute to cessation of milk secretion implying the importance of glucose in milk secretion (Rook et al. 1965).Thus, a reduced blood perfusion to the mammary gland and less glucose uptake by the mammary cells may occur in mastitis. In turn, depressed milk yield is seen which could explain one of the important clinical sign in mastitis.

5.4.3 Protein Metabolism Ruminant animals require two types of digestible proteins; the one which escapes rumen degradation which is digested in the small intestine to be used by the animals itself and the one is a degradable protein in the rumen used by the micro-organisms to synthesize microbial protein. Proteins are important sources for maintenance of vital functions, reproduction, growth and lactation. Within the mammary gland, large amounts of amino acids are needed to synthesize milk proteins. Amino acids can also be converted into other non-essential amino acids or oxidized into energy in the mammary gland (Wattiaux and Karg 2004).

In this study, the six protein metabolic pathways with greatest change were chosen for discussion and in each metabolic pathways, six metabolites were identified that increased or decreased in mastitis compared to control animals. However, breakdown of milk proteins such as casein is most likely to be responsible for much of the amino acid and di-, tripeptides formed in this study.

Amongst these metabolites, it is apparent that methionine (Met) is increased in both mastitic groups. As mentioned previously, amino acids in milk are absorbed from blood but also synthesized and degraded in the mammary gland (Bequette et al. 1998). According to Pinotti et al. (2002), healthy dairy ruminants produce large amounts of methionine or a combination of Met and lysine (Lys) (Schwab et al. 1992). But in the present study, no increase in free Lys was observed in mastitic groups. These amino acids are the first limiting amino acids for milk protein synthesis and Met particularly is required for transmethylation reactions and milk protein synthesis and also able to stimulate both Chapter 5, 269 milk and milk fat productions. However, the level of amino acids in milk is greatly influenced by the quality and quantity of amino acids in basal diets (Klein et al. 2010).

Glycine (Gly) level is increased in milk from both mastitic groups in this present study and this increased has been observed during the early lactation due to increased Gly metabolism during this period (Klein et al. 2010). Phosphocholine levels showed a positive correlation (r=0.81) with the levels of Gly which is also increased during early stage of lactation (Klein et al. 2010). In this present study, phosphocholine is also increased in mastitic milk samples. Phosphocholine acts as a precursor of Gly and serves as its storage form within the cytosol. It is therefore possible that the increased levels of both Gly and phosphocholine found in this study are due to mastitic milk collected at an early stage of lactation and that the ability of mammary gland to synthesize these amino acids are not impaired by udder infection.

High levels of both tyrosine and isoleucine amino acids are seen in the present study in mastitic milk samples. These amino acids are seen to increase in organic milk compared to conventional whole milk as studied by Boudonck et al. (2009). High tyrosine levels in mastitic milks can be explained by the extensive proteolysis in milk. Proteolysis can result in accumulation of small peptides which contribute to the bitterness in the milk. According to Murphy et al. (1989), casein proteolysis products can be measured by the determination of tyrosine level. Thus, these authors agreed that high values on tyrosine during infection period are due to high milk protease activity during mastitis.

Bacterial fermentation in the milk can produced milk products which contained considerably amounts of D-amino acids (Man and Bada 1987) and bacteria can be regarded as a biological source of D-amino acids (D-AA). Study by Pohn et al. (2009) revealed greater increases of free amino acids are seen in milk with different scores of CMT especially Ile, Ala, Asp, Pro and Leu. This is in accordance with the present study in which Ile, Pro and Leu were increased in mastitic milk samples. Study by Pohn et al. (2009) also observed high free D-AA as CMT score increased for all nine free D-AA (D- Asp, D-Ser, D-Glu, D-Pro, D-Ala, D-Val, D-allo-Ile, D-Leu and D-Lys). In this present study, D-Pro and D-Cys did increase in mastitic milk samples. Thus, it can be concluded the presence of D-AA content is associated with mastitis and related to the bacterial activity in the udder. However, the MS analysis cannot differentiate D- from L- amino aicd as they have identified m/z and retention time, so that identification as D- amino acid is not definitive. Chapter 5, 270

It can be concluded that increase in many amino acids in mastitic milk samples is due to proteolytic activity caused by proteases from pathogens and also from immune cells such as plasmin caused by the presence of microorganisms during udder infection. End products of major milk proteins degradation by proteases contribute to a wide-range of amino acids present in the mastitic milk samples. Thus, certain amino acids can be used to indicate the activity of pathogenic bacteria which lead to inflammation of mammary gland.

In conclusion, there was an increase in identification of di-and tripeptides in both infections with particular tripeptides being associated with either of the bacteria, though with S. aureus infection causing more of these peptides to be released. Differences that present between bacterial causative agents could be either due to differences in the specificity of action of the bacterial proteases or that the bacteria have differing efficiencies in absorbing the peptides after they are released by hydrolysis. Amongst other metabolites changes in lipid, nucleotide and carbohydrate metabolism have been identified, there were also non-host and non-pathogen metabolites identified such as tentoxin, ricinine, quinate, toluene and others and it was possible that the metabolites are derived from therapeutic reagents though mis-classification is also possible.

To differentiate between E. coli and S. aureus infected milk samples, certain metabolites may have potential as biomarkers such as pyridoxine, N-acetyl-L-glutamate, isopropylmalate and pyruvate. These metabolites were seen in E. coli infected milk samples as compared with S. aureus infected milk samples. The majority of increased metabolites in S. aureus infected milk samples were comprised of di- or tri-peptides such as Gln-Ser-Ser, Leu-Asn-Tyr, Leu-Thr and others.

Mastitis had an effect on lipid, protein and carbohydrate metabolic pathways. In lipid metabolic pathway, there was an increased in arachidonic acid metabolites in mastitis. As for protein metabolic pathway, arginine metabolites were seen increased in mastitis whilst galactose metabolites from carbohydrate metabolic pathway were increased in mastitic milk samples.

From this study, it can be concluded that metabolomic analysis has successfully studied the presence of important metabolites or even determined the changes of metabolites that occurred in milk samples of dairy cows during the course of intramammary inflammation. Further analysis is needed with more milk samples being studied to confirm the status of these important metabolites and also to assess their appearance in subclinical as well as the clinical mastitic samples studied in this investigation. Chapter 6, 271

Chapter VI. General Discussion

The aims of this thesis were outlined in chapter 1 and the relevant conclusions have been detailed in the appropriate chapters. This chapter will summarise the important and significant findings and highlight the areas which have been recognised as worthy of future research.

Despite the numerous researches that have been undertaken to control bovine mastitis, this disease still poses one of the most prevalent and costly diseases in dairy cows. The negative impacts of bovine mastitis include reduction in milk production and quality because of changes in milk composition which in turn affects the economy of dairy industry. Other factors such as costs of treatment, veterinary services, diagnostics and culling were also contribute to the detrimental effect on dairy industry worldwide (Halasa et al. 2007).

In chapter 1, various traditional and common bovine mastitis diagnostic methods have been discussed. Somatic cell counts (SCC) is still widely used as one of the most useful diagnostic techniques to detect the level and occurrence of bovine mastitis especially for bovine subclinical mastitis (Dohoo & Meek 1982; Schukken et al. 2003). However, in order to avoid false-positive or false-negative results, several factors should be taken into consideration before any conclusion can be made. These factors include mastitis causing pathogens, number of infected quarters, previous mastitis history, age, stage of lactation, milking frequency as well as the type of samples taken for SCC analysis (quarter, composite or bulk tank).

Other important diagnostic methods that have been discussed in detecting the presence of bovine mastitis are California Mastitis Test (CMT), enzyme levels, electrical conductivity (EC) and infrared thermography (IRT). CMT which was developed by Schalm & Noorlander (1957) is the main cow-side test used to diagnose bovine mastitis and is based on the formation of gel formation which reflects the cell count of the milk (Barnum & Newbould 1961). Previous studies have been shown the relationship of CMT results with the leukocyte content of milk, the level of subclinical mastitis as well as bacteriological analyses (Barnum & Newbould 1961; Daniel et al. 1966; Wesen et al. 1967).

Several types of enzymes may be useful as mastitis indicators but one enzyme that is considered reliable and sensitive is N-acetyl-β-D-glucosaminidase (NAGase) (Kitchen 1976). The use of both SCC and NAGase provides valuable interpretation for the diagnosis Chapter 6, 272 of bovine mastitis so that the level and extent of infection can be controlled (Pyörälä & Pyörälä 1997). Other important enzymes include β-glucuronidase, milk antitrypsin, L- lactate dehydrogenase (LDH) and myeloperoxidase (MPO).

The use of EC has been one of the most popular methods in diagnosing bovine mastitis. The concentration of NaCl is often measured as milk electrical conductivity and many studies have been performed to assess the ability of EC in diagnosing bovine mastitis particularly subclinical mastitis (Linzell et al. 1974; Fernando et al. 1982, 1985).

Finally, another popular bovine mastitis diagnostic method is IRT. It has been shown that IRT has strong correlation with CMT and a high predictive diagnostic ability making it a sensitive tool in detecting early onset of subclinical mastitis (Berry et al. 2003; Polat et al. 2010) although several cow factors should be consider before analysing the reliability of IRT in diagnosing bovine mastitis.

However, these traditional methods of bovine mastitis diagnosis may have their technical drawbacks such as less sensitive, less specific, lacking in speed, very labour-intensive and expensive. Their results can also be mis-interpreted which then lead to false positive or negative (Chagunda et al. 2006).

Recently, much research interest has been focused on analysing the presence of bovine mastitis biomarkers that can be used to predict the progress and outcome of a disease. Biomarkers are important analytes which can be used to predict changes in any biological and pathological processes so that any disease states can be revealed (Biomarkers Definitions Working Group 2001). Because of the limitations in traditional methods of diagnosing bovine mastitis that had been discussed previously (Chagunda et al. 2006; Le Roux et al. 2003; Labohm et al. 1998; Nielen et al. 1995), the discovery of specific mastitis biomarkers is needed for an early diagnosis of bovine mastitis. Thus, several bovine mastitis biomarkers have been proposed including the use of combinations of several bovine mastitis indicators such as enzymes, SCC, prostanoids, bacteriological culture and others (Mattila et al. 1986; Zank & Schlatterer 1998; Barta et al. 1990; Baeker et al. 2002; Bannikov et al. 2010).

Two important APPs that are known to increase in response to disease are SAA and Hp (Conner et al. 1986; Eckersall et al. 2001). Hp and SAA are also known to rise substantially in serum or milk as indicators of bovine mastitis (Pyörälä et al. 2011). In chapter 2, the reliability of both SAA and Hp was assessed by validating and optimising these APPs assays with the use of bovine milk samples. Chapter 6, 273

The aim of chapter 2 was to develop sandwich ELISA by using rabbit anti-bovine antibody to measure Hp level in bovine milk and used commercialized PhaseTM Range SAA Assay to quantify the level of milk SAA so that the concentrations of the APPs in different types of milk samples could be compared statistically. To group milk samples into healthy, subclinical and clinical mastitis, the estimation of SCC and bacteriological culture were used before APP assays were performed in order to confirm the disease status. It was shown in this chapter that Hp concentrations showed to be statistically increased in clinical mastitic samples when compared between healthy and clinical mastitic cases as well as between subclinical and clinical mastitic cases. However, Hp showed no significant difference between healthy and subclinical mastitic milk samples (p>0.05). Several factors were discussed which may contribute to this reason which include different types of cytokines produced during bovine mastitis, less Hp production during subclinical mastitis, insensitive immunoassay to detect a slight elevation of Hp concentrations in subclinical mastitic milk samples as well as different types of pathogen virulence and host responses.

Milk serum amyloid A3 (M-SAA3) has also been shown to have an important role as biomarker in bovine mastitis (Eckersall et al. 2001; Grönlund et al. 2003, 2005; Nielsen et al. 2004; Pyörälä et al. 2011). This study successfully confirmed the use of M-SAA3 in diagnosing bovine mastitis in which M-SAA3 concentrations were significantly different between healthy, subclinical and clinical mastitic milk samples (p<0.05).

SAA levels showed a relatively strong relationship with Hp levels further confirming that SAA and Hp can be used together as good mastitis indicator. It has also been shown that these APP measurements can be performed using milk samples instead of serum/plasma samples for an ease of sampling and sample preparation.

Somatic cell counts (SCC) were measured in both healthy and subclinical mastitic milk samples and the relationship between SCC values and APP measurements was determined. SCC measurement was significantly different between healthy and subclinical mastitic samples (p<0.05) which proved the reliability of SCC to diagnose subclinical mastitis. The correlation between SCC and APP showed positive relationship although the relationship was not strong. The correlation between SCC and SAA gave a medium relationship with the value of r was 0.438 while here was no correlation between SCC and Hp as the value of r was 0.08. The reason for this difference could be that these correlations were calculated based on all milk samples regardless of the type of milk samples (control vs. infected) and various cut-off values for SCC have been suggested depending on several factors (countries, breeds, types of milk samples) (Mattila et al. 1986; Holdaway et al. Chapter 6, 274

1996). The use of log transformed SCC can be considered when analysing their relationship to achieve good correlation between APP and SCC.

The application of advances in proteomics has been a major recent area of research in studying the protein changes in bovine milk during mastitis. In chapter 3, several proteomics methods were applied, such as 1DE, 2DE, DiGE and MALDI-MS analyses, to investigate the changes of milk proteome during the presence of intramammary inflammation. Milk samples from subclinical and clinical mastitic cases were used in comparison with healthy milk samples as comparison. It was hoped that by applying these methods, discovery of new biomarkers of bovine mastitis can be achieved.

Prior to the use of proteomic methods, SCC and APP measurements were used to confirm the disease status of the milk samples investigated. In initial studies rennet coagulation was performed to remove the majority of milk casein in order to examine changes in low abundance proteins. In the first phase of this study, subclinical mastitic milk samples were analysed in comparison with healthy milk samples using 1D gel electrophoresis and the protein bands developed on the gel were analysed by the MALDI-MS for protein identification. Unfortunately, no clear difference was seen on the gel although lower intensity bands but with high molecular weight were not successfully identified by the MALDI-MS analysis.

Because no clear difference was observed on the gel pattern of 1D gel electrophoresis, 2D gel electrophoresis was applied to similar samples. Unfortunately, there was again no observable difference in low abundance proteins between pooled subclinical mastitic and pooled healthy milk samples. However, it was evident that albumin was not present in healthy milk samples which further supported the idea that albumin concentrations was increased during bovine mastitis as a result of either an increase in leakage from serum into the mammary gland or in albumin expression in the mammary gland.

Due to these limitations, clinical mastitic milk samples were analysed by using DiGE which enables the visualization of multiple protein samples in one 2D gel. However, the only findings observed were increases in albumin concentration and decreased whey protein concentrations (β-LG and α-LA) in clinical mastitic milk samples as compared with healthy milk samples. Due to lack of significant findings in this chapter, non-gel proteomic approaches were recommended to continue the investigation of bovine mastitis biomarkers. Chapter 6, 275

Peptidomic approach was applied next to investigate the presence of polypeptides in bovine milk that may be used to discriminate normal from bovine mastitic milk samples and also to determine if peptide biomarkers could discriminate between mastitis caused by two different types of bacterial pathogens. It was hoped that by using this sensitive and high-throughput methodology, significant results can be achieved and lead to the discovery of important and significant bovine mastitis biomarkers by employing CE-MS and LC- MS/MS analyses.

Milk samples used in this study were from clinical mastitic cases, caused by two common bacterial pathogens which were E. coli and S. aureus, and were assessed in comparison to healthy milk samples. Two different types of comparison were analysed to determine the ability of peptidomics to compare between healthy and clinically infected milk samples as well as the ability of peptidomics to compare between E. coli infected and S. aureus infected milk samples. SCC and APP measurements were used as previously to confirm the disease status of the milk samples used.

From CE-MS analysis, polypeptide fingerprinting was performed for different groups of milk samples to generate disease models. However, these models need to be validated so that the status of each marker in each group (control, infected, E. coli or S. aureus) can be confirmed and may eventually used to determine the presence of bovine mastitis or even predict the type of bacterial pathogens that caused bovine mastitis.

A total of 45 discriminatory polypeptides were identified by CE-MS analysis with 31 polypeptides used to discriminate healthy from mastitis cases and 14 polypeptides discriminating E. coli from S. aureus infected cases. These polypeptides were then further analysed for their amino acid sequence by LC-MS/MS, although not all polypeptides were successfully sequenced. The reason was because the only parameters used to link the peptides obtained in both systems (CE-MS and LC-MS/MS) were peptide mass and the number of basic amino acids. Thus, mass ambiguities occurred which complicate the assignment of sequences obtained from these two different methods.

However, a total of 12 sequenced polypeptides were identified using UniProtKB database. Amongst the proteins identified from the peptide sequences were α S1-casein, β-casein, serum amyloid A protein (SAA), glycosylation-dependent cell adhesion molecule 1 (GlyCAM-1) and fibroblast growth factor binding protein (FGFBP). Although GlyCAM-1 was characterized as one of the host defence mechanism proteins (Boehmer et al. 2010; Smolenski et al. 2007), it was interesting to see the presence of these peptides from GlyCAM-1 in healthy milk samples. Peptides from SAA were also seen in both E. coli and Chapter 6, 276

S. aureus infected milk samples but they were not able to discriminate E. coli infected milk samples from S. aureus infected milk samples. The results from this chapter provided an initial insight on how milk peptide profiles can change as a result of bovine mastitis. Further improvements have been discussed but it was confirmed that peptidomic methodology able to identify significant biomarkers in bovine mastitis.

Finally, a metabolomic approach was applied to further investigate the presence of important metabolites that can be used to discriminate normal from mastitic milk samples by the use of LC-MS/MS analysis. As previously, SCC and APP measurements were performed to confirm the disease status of the samples used. Individual metabolites were analysed in both positive and negative ionisation modes in each different types of milk sample. Metabolites which were absent in control milk samples but present in both infected groups (E. coli and S. aureus) were also examined in both ionisation modes. Among the many metabolites detected, a variety of tri-peptides were seen with higher intensities in S. aureus group compared with E. coli group regardless of the type of ionisation modes. The relative change in intensity for each identified metabolites between two infected groups was determined and Mann-Whitney test was performed to show the significant difference in each of the categories. Amongst the identified metabolites which were significantly increased in the E. coli infected group regardless of the type of ionisation modes were 3- nitrotyrosine, deoxyguanosine and 3-methoxy-4-hydroxylphenyl. Whilst Leu-Arg, Ser-His and Lys-Ala-Gly were amongst the identified metabolites found significantly increased in S. aureus infected group.

Metabolic pathways were also analysed to allow an investigation of the metabolites found in different group of milk samples according to their corresponding metabolic pathways. In lipid metabolic pathway, arachidonic acid metabolites were seen increased in mastitis while in protein metabolic pathway, arginine metabolites were found to be increased in mastitis. Finally, galactose metabolites from carbohydrate metabolic pathways were increased in mastitic milk samples.

In chapter 4 and 5, several significant findings were found which could be proposed as bovine mastitis biomarkers. It was also apparent that some of the milk indicators could be applied to differentiate clinical mastitis caused by different type of bacterial pathogens (E. coli and S. aureus). This would be of especial value in allowing the appropriate antibacterial treatment. However, further experiments with more milk samples should be performed to validate and confirm these results so that they can be used to diagnose bovine mastitis. Furthermore it is important to investigate the differentiation of subclinical and Chapter 6, 277 clinical bovine mastitis, so that subclinical mastitic milk samples should be used in both peptidomic and metabolomics methodologies in comparison with clinical mastitic and healthy milk samples as it was apparent that the use of proteomic methodology in this study was not able to differentiate between subclinical and clinical bovine mastitis. It is also hoped that changes in bovine milk proteome, peptidome or metabolome can be perceived efficiently.

In summary of this thesis, with a prior knowledge of known acute phase proteins during bovine mastitis, various ‘-omics’ studies such as proteomics, peptidomics and metabolomics can be applied to study the presence of other lower abundance proteins, peptides and metabolites which may serve as important mastitis biomarkers. 278

Appendices

Appendix Chapter 3

1. Preparation of 12% PAGE gels

The monomer solution was prepared by mixin APS. The mixture was degassed for 15 minutes.

Percent Gel DDI water (ml) 30% Degassed  Gel 10% w/v SDS Acrylamide/Bis buffer (ml) (ml) (ml) 12% 3.4 4.0 2.5 0.1  Resolving Gel Buffer – 1.5M Tris-HCl, pH 8.8

 Stacking Gel Buffer – 0.5M Tris-HCl, pH 6.8

Immediately prior to pouring the gel, for 10 ml monomer solution, these solutions were added:

Resolving Gel: 50μl 10% APS and 5μl TEMED

Stacking Gel: 50μl 10% APS and 10μl TEMED

The mixed solution was gently swirled to initiate polymerization

2. Bradford Assay The Bradford reagent which consists of Brillian Blue G in phosphoric acid and methanol as well as Protein Standard (BSA) Solution (2mg/ml) were purchased from Sigma-Aldrich.Inc.

a) Protein Standard preparation

Bovine serum albumin at 2mg/ml was used as the protein standard in performing Bradford assay. The standard was further diluted using MilliQ water into 1.5mg/ml, 1.0mg/ml and 0.5mg/ml to produce a standard curve.

b) Milk samples preparation 279

Milk samples were used from both healthy and mastitic cows. All samples were kept at -20°C before analysis. After thawed, all samples were further diluted in 1:2, 1:4, 1:6 and 1:8 using MilliQ water. This dilution procedure was not necessary to perform in all assays.

c) Assay

250μl of Bradford reagent was added into 5μl of each standard and sample at different dilutions. The sample was incubated for a few minutes at room temperature and the absorbance was measure at 595nm using [name].

d) Protein Concentration Analysis

The net absorbance values against the protein concentration of each standard were plotted automatically after the measurement. The protein concentration of each sample was determined by comparing the Net

A595 values against the standard curve.

3. Lysis buffer 6M urea

2M thiourea

4% CHAPS

25mM tris base

4. Equilibration buffer I pH8.8, 1.5M Tris HCl

216.21 g urea

180ml glycerol

12g SDS

100mg DTT

5. Equilibration buffer II pH8.8, 1.5M Tris HCl

216.21 g urea

180ml glycerol

12g SDS 280

250mg iodoacetamide

6. DiGE Lysis Buffer 7M urea

2M Thiourea

4% CHAPS

25mM Tris HCl

Appendix Chapter 4

1. Preparation of Standards for microplate assay

Standard BSA standard Water or buffer Final protein concentration Standard A 75 µL stock 225 µl 500 µg/ml (2 mg/ml) Standard B 200 µL of A 200 µl 250 µg/ml

Standard C 200 µL of B 200 µl 125 µg/ml

Standard D 80 µL of C 120 µl 50 µg/ml

Standard E 60 µL of D 140 µl 15 µg/ml

Standard F 60 µL of E 120 µl 5 µg/ml

Standard G 80 µL of F 120 µl 2 µg/ml

Blank 0 200 µl 0 µg/ml

2. Preparation of BC Assay reagent

Prepare the required amount of BC Assay reagent by adding:

1 part of reagent B to 50 parts of reagent A.

3. Components for sample preparation 281

1. Disposable de-salting PD10-columns (GE Healthcare Biosciences AB, Bedford, USA). 2. Centristart ultrafilter tubes 20 kD, 0.62 cm2 tubes (Sartorius Stedim Limited, UK). 3. BCA protein quantification kit (Uptima, Interchim, Montluçon, France). 4. Lyophilisation Christ Speed-Vac RVC 2-18/Alpha 1-2 (Christ, Osterode am Harz, Germany).

4. Components for CE-MS analysis

1. P/ACE MDQ capillary electrophoresis system (Beckman Coulter, Crea, CA, USA)

2. Micro-TOF (time-of-flight) mass spectrometer (Bruker Daltonics, Bremen, Germany).

3. Electro-ionisation sprayer (Agilent Technologies, Santa Clara, USA) (see Note 3).

4. Capillary 90 cm, 50 µm I.D. fused silica capillary non-coated (New Objective, Woburn, USA)

5. Cole-Parmer single-syringe infusion pump (Cole-Parmer Instrument Company, USA).

6. Hamilton syringe 2.5 ml

5. Software for Data processing and Statistical Analysis a) Software for peak detection: MosaiquesVisu (Biomosaiques, Hannover, Germany). b) Commercial statistical packages: SAS (www.sas.com) and MedCal (www.medcalc.be). c) Multitest R-package available at www.bioconductor.org d) SVM-based program for multiparametric data classification: MosaCluster (Biomosaiques, Hannover, Germany).

282

List of Reference

Addeo, P., Moio, L., Chianese, L., Stingo, C. & Di Luccia, A. (1990). Improved procedure for detecting bovine and ovine milk mixtures in cheese by isoelectric focusing of para-κ-casein. Milchioissenschafl 45: 221-224.

Aebersold, R. and Mann, M. (2003). Mass spectrometry-based proteomics. Nature 422: 198-207.

Affolter, M., Grass, L., Vanrobaeys, F., Casado, B. and Kussmann, M. (2010). Qualitative and quantitative profiling of the bovine milk fat globule membrane proteome. Journal of Proteomics 73: 1079-1088.

Åkerstedt, M., Björk, L., Waller, K.P. and Sternesjö, Å. (2006). Biosensor assay for determination of haptoglobin in bovine mastitis. Journal of Dairy Research 73: 299-305.

Åkerstedt, M., Waller, K.P. and Sternesjö, Å. (2007). Haptoglobin and serum amyloid A in relation to the somatic cell count in quarter, cow composite and bulk tank milk samples. Journal of Dairy Research 74: 198-203.

Åkerstedt, M., Waller, K.P. and Sternesjö, Å. (2009). Haptoglobin and serum amyloid A in bulk tank milk in relation to raw milk quality. Journal of Dairy Research 76: 483-489.

Åkerstedt, M., Waller, K.P., Larsen, L.B., Forsbäck, L. and Sternesjö, Å. (2008). Relationship between haptoglobin and serum amyloid A in milk and milk quality. International Dairy Journal 18: 669-674.

Aldo-Benson, M.A. and Benson, M.D. (1982). SAA suppression of immune response in vitro: Evidence of an effect on T-cell macrophage interaction. Journal of Immunology 128: 2390-2392.

Alsemgeest, S.P.M., van’t Klooster, G.A.E, van Miert, A.S.J.P.A.M., Hulskamp-Koch, C.K. and Gruys, E. (1996). Primary bovine hepatocytes in the study of cytokine induced acute-phase protein secretion in vitro. Veterinary Immunology and Immunopathology 53: 179-184.

Álvarez-Sánchez, B., Priego-Capote, F. & Luque de Castro, M.D. (2010). Metabolomics analysis I. Selection of biological samples and practical aspects preceding sample preparation. Trends Anal. Chem. 29: 111–119.

Anderson, K. L., Kindahl, H., Petroni, A., Smith, A.R. and Gustafsson, B.K. (1985). Arachidonic acid metabolites in milk of cows during acute coliform mastitis. Am. J. Vet. Res. 46:1573. 283

Anderson, M. and Needs, E. C. (1983). Serum lipoprotein stimulation of lipolysis and its relevance to free fatty acid development in bovine milk. Journal of Dairy Research 50: 309-319.

Anderson, N.G., Powers, M.T. and Tollaksen, S.L. (1982). Proteins of human milk. I. Identification of major components. Clinical Chemistry 28: 1045-1055.

Anderson, N.L. and Anderson, N.G. (2003). The human plasma proteome: history, character and diagnostic prospects. Molecular Cell Proteomics 1: 845-867.

Aristoteli, L.P., Molloy, M.P. and Baker, M.S. (2007). Evaluation of endogenous plasma peptide extraction methods for mass spectrometric biomarker discovery. Journal of Proteome Research 6: 571-581.

Atroshi, F., Parantainen, J., Kangasniemi, R. and Osterman, T. (1987). Milk prostaglandins and electrical conductivity in bovine mastitis. Veterinary Research Communications 11: 15-22.

Atroshi, F., Parantainen, J., Sankari, S. and Usterman, T. (1986). Prostaglandins and glutathione peroxidase in bovine mastitis. Res. Vet. Sci. 40: 361-363.

Atroshi, F., Rizzo, A., Osterman, T. and Parantainen, J. (1989). Free fatty acids and lipid peroxidation in normal and mastitic bovine milk. Journal of Veterinary Medicine Series A. 36: 321-330.

Azzara, C.D. and Dimick, P.S. (1985). Lipoprotein lipase activity of milk from cows with prolonged subclinical mastitis. Journal of Dairy Science 68: 3171-3175.

Badolato, R., Johnston, J.A., Wang, J.M., McVicar, D., Xu, L.L., Oppenheim, J.J. and Kelvin, D.J. (1995). Serum amyloid A induces calcium mobilization and chemotaxis of human monocytes by activating a pertussis-toxin-sensitive signalling pathway. Journal of Immunology 155: 4004-4010.

Badolato, R., Wang, J.M., Murphy, W.J., Lloyd, A.R., Michiel, D.F., Bausserman, L.L., Kelvin, D.J. and Oppenheim, J.J. (1994). Serum amyloid A is a chemoattractant: Induction of migration, adhesion and tissue infiltration of monocytes and poylmorphonuclear leukocytes. The Journal of Experimental Medicine 180: 203-209.

Baeker, R., Haebel, S., Schlatterer, K. and Schlatterer, B. (2002). Lipocalin-type prostaglandin D synthase in milk: a new biomarker for bovine mastitis. Prostaglandins and Other Lipid Mediators 67: 75-88.

Baggerman, G., Cerstiaens, A., De Loof, A. and Schoofs, L. (2002). Peptidomics of the larval Drosophila melanogaster central nervous system. J. Biol. Chem. 277:40368–40374. 284

Baggerman, G., Verleyen, P., Clynen, E., Huybrechts, J., De Loof, A. and Schoofs, L. (2004). Peptidomics. Journal of Chromatography B-Analytical Technologies in the Biomedical and Life Sciences 803: 3-16.

Baldwin, R.L. and Cheng, W. (1969). Metabolite changes associated with initiation and maintenance of lactation in rats and cows. Journal of Dairy Science 52: 523-528.

Ball H.A., Cook J.A., Wise W.C. and Halushka P.V. (1986). Role of thromboxane, prostaglandins and luekotrienes in endotoxic and septic shock. Intensive Care Medicine 12: 116-126.

Banderas, M.G., Laborde, C.M., Posada, M., de la Cuesta, F., Zubiri, I., Vivanco, F. and Alvarez- Llamas, G. (2011). Metabolomic profiling for identification of novel potential biomarkers in cardiovascular diseases. Journal of Biomedicine and Biotechnology 2011: 1-9.

Bannerman, D.D. (2009). Pathogen-dependent induction of cytokines and other soluble inflammatory mediators during intramammary infection of dairy cows. Journal of Animal Science 87: 10-25.

Bannikov, G.A., Hinds, C.A., Rajal-Schultz, P.J., Premanandan, C., Rings, D.M. and Lakritz, J. (2010). Serum haptoglobin-matrix metalloproteinase 9 (Hp-MMP 9) complex as a biomarker of systemic inflammation in cattle. Vet Immunol Immunopath 139: 41-49.

Bantscheff, M., Schirle, M., Sweetman, G., Rick, J. and Kuster, B. (2007). Quantitative mass spectrometry in proteomics: a critical review. Anal. Bioanal. Chem. 389: 1017–1031.

Barber, E.F. and Cousins, R.J. (1988). Interleukin-1- Stimulated induction of ceruloplasmin synthesis in normal and copper-deficient rats. The Journal of Nutrition 118: 375-381.

Barkema, H.W., Deluyker, H.A., Schukken, Y.H. and Lam, T.J.G.M. (1999). Quarter-milk somatic cell count at calving and at the first six milkings after calving. Prev. Vet. Med. 38: 1-9.

Barnum, D.A. and Newbould, F.H.S. (1961). The use of the California Mastitis Test for the detection of bovine mastitis. The Canadian Veterinary Journal 2:83-90.

Barta, O., Barta, V.D., Crisman, M.V. and Akers, R.M. (1990). Lymphocyte blastogenesis inhibition by milk whey as an indicator of mastitis. Journal of Dairy Science 73: 2112-2120.

Barth, K. (2000). Basic investigations to evaluate a highly sensitive infrared-thermograph-technique to detect udder inflammation in cows. Milchwissenschaft 55: 607–609.

Bartsch, B.D., Beck, C.G. and Hehir, A.F. (1981). Mastitis and the fatty acid composition of milk fat. Aust. J. Dairy Technol. 35: 93-94. 285

Baseler, M.W. and Burrell, R. (1983). Purification of haptoglobin and its effect on lymphocyte and macrophage responses. Inflammation. 7: 387-400.

Bastian, E.D., Hansen, K.G. and Brown, R.J. (1991). Activation of plasmin with urokinase in ultrafiltered milk for cheese manufacture. Journal of Dairy Science 74: 3669-3676.

Baumann, H. & Gauldie, J. (1994). The acute phase response. Immunol. Today 15: 74–80.

Baumann, S., Ceglarek, U., Fiedler, G.M., Lembcke, J., Leichtle, A. and Thiery, J. (2005). Standardized approach to proteome profiling of human serum based on magnetic bead separation and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Clin Chem. 51:973–80.

Beavis, R. and Fenyo, D. (2000). Proteomics: A Trends Guide, 22.

Benditt, E.P. and Eriksen, N. (1977) Amyloid protein SAA is associated with high-density lipoprotein from human serum (apolipoprotein) Proceedings of the National Academy of Sciences of the United States of America 74: 4025-4028.

Benditt, E.R. and Meek, R.L. (1989). Expression of the third member of the serum amyloid A gene family in mouse adipocytes. J. Exp. Med. 169: 1841–1846.

Benson, M.D. and Aldo-Benson, M. (1979). Effect of purified protein SAA on immune response in vitro: Mechanisms of suppression. The Journal of Immunology 122: 2077-2082.

Bequette, B.J., Backwell, F.R.C. and Crompton, L.A. (1998). Current concepts of amino acids and protein metabolism in the mammary gland of the lactating ruminant. Journal of Dairy Science 81: 2540-2559.

Beranova-Giorgianni, (2003). Proteome analysis by two-dimensional gel electrophoresis and mass spectrometry: strengths and limitations. Trends in Analytical Chemistry 22: 273-281.

Bernabucci, U., Ronchi, B., Lacetera, N. and Nardone, A. (2005). Influence of body condition score on relationships between metabolic status and oxidative stress in periparturient dairy cows. J. Dairy Sci. 88: 2017-2026.

Berning, L.M. and Shook, G.E., (1992). Prediction of mastitis using milk somatic cell count, N- acetyl-β-D-glucosaminidase and lactose. Journal of Dairy Science 75: 1840-1848.

Berry, R. J., Kennedy, A.D., Scott, S.L., Kyle, B.L. and Schaefer, A.L. (2003). Daily variation in the udder surface temperature of dairy cows measured by infrared thermography: Potential for mastitis detection. Can. J. Anim. Sci. 83:687–693. 286

Bertram, H.C., Kristensen, N.B., Malmendal, A., Nielsen, N.C., Jensen, S.K. and Harmon, D.L. (2004). An NMR-based metabolomic approach to assess metabolism in splanchnic tissues of steers. Journal of Animal and Feed Sciences 13: 295-298.

Bertram H.C., Petersen B.O., Duus J. Ø., Larsen M., Raun B.M.L. and Kristensen N.B. (2009). Proton nuclear magnetic resonance spectroscopy based investigation on propylene glycol toxicosis in a Holstein cow. Acta Veterinaria Scandinavica 51: 25.

Bhikane, A.V. & Kawitkar S.B. (2000). Handbook for Veterinary Clinician. Venkateh Books. Udgir, India.

Bianchi, L., Puglia, M., Landi, C., Matteoni, S., Perini, D., Armini, A., Verani, M., Trombetta, C., Soldani, P., Roncada, P., Greppi, G., Pallini, V. and Bini, L. (2009). Solubilization methods and reference 2-DE map of cow milk fat globules. Journal of Proteomics 72: 853-864.

Biggadike, H.J., Ohnstad, I., Laven, R.A. and Hillerton, J.E. (2002). Evaluation of measurements of the conductivity of quarter milk samples for the early diagnosis of mastitis. Veterinary Record 150: 655-658.

Biomarkers Definitions Working Group. (2001). Biomarkers and surrogate endpoints: Preferred definitions and conceptual framework. Clinical Pharmacology and Therapeutics 69: 89-95.

Boehmer, J.L., Bannermann, D.D., Shefneck, K. and Ward, J.L. (2008). Proteomic analysis of differentially expressed proteins in bovine milk during experimentally induced Escherichia coli mastitis. Journal of Dairy Science 91: 4206-4218.

Boehmer, J.L., DeGrasse, J.A., McFarland, M.A., Tall, E.A., Shefcheck, K.J., Ward, J.L. and Bannerman, D.D. (2010). The proteomic advantage: Label-free quantification of proteins expressed in bovine milk during experimentally induced coliform mastitis. Veterinary Immunology and Immunopathology 138: 252-266.

Boehmer, J.L., Ward, J.L., Peters, R.R., Shefneck, K.J., McFarland, M.A. and Bannermann, D.D. (2010). Proteomic analysis of the temporal expression of bovine milk proteins during coliform mastitis and label-free relative quantification. Journal of Dairy Science 93: 593-603.

Bogin, E., Ziv, G., Avidor, J., Rivetz, B., Gordin, S. & Saran, A. (1977). Distribution of lactate dehydrogenase isoenzymes in normal and inflamed bovine udder and milk. Research in Veterinary Science 22:198–200. 287

Boos, K.S. and Grimm, C.H. (1999). High-performance liquid chromatography integrated solid-phase extraction in bioanalysis using restricted access precolum n packings. Trends in Analytical Chemistry 18: 175-180.

Boosman, R., Niewold, T.A., Mutsaers, C.W.A.A. and Gruys, E. (1989). Serum amyloid A concentrations in cows given endotoxin as an acute-phase stimulant. Am. J. Vet. Res. 50: 1690–1694.

Bottero, M.T., Civera, T., Nucera, D., Rosati, S., Sacchi, P. and Turi, R.M. (2003). A multiplex polymerase chain reaction for the identification of cows’, goats’ and sheep’s milk in dairy products. International Dairy Journal 13: 277-282.

Bottoms, G. D. and Adams, H.R. (1992). Involvement of prostaglandins and leukotrienes in the pathogenesis of endotoxemia and sepsis. J. Am. Vet. Med. Assoc. 200:1842–1848.

Bouchard, L., Blais, S., Desrosiers, C., Zhao, X. and Lacasse, P. (1999). Nitric oxide production during endotoxin-induced mastitis in the cow. Journal of Dairy Science 82: 2574-2581.

Boudonck, K.J., Mitchell, M.W., Wulff, J. and Ryals, J.A. (2009). Characterization of the biochemical variability of the bovine milk using metabolomics. Metabolomics 5: 375-386.

Boutet, P., Bureau, F., Degand, G. and Lekeux, P. (2003). Imbalance between lipoxin A4 and leukotriene B4 in chronic mastitis-affected cows. Journal of Dairy Science 86: 3430-3439.

Boyce, J.D., Cullen, P.A., Nguyen, V., Wilkie, I. and Adler, B. (2006). Analysis of the Pasteurella multocida outer membrane sub-protoeme and its response to the in vivo environment of the natural host. Proteomics 6: 870-880.

Bramley, A.J. and Schultze, W.D. (1991). Effect of milking without pulsation on teat duct colonization with Streptococcus agalactiae and penetrability to endotoxin. Journal of Dairy Science 74: 2982-2988.

Bremner, K.C. (1964). Studies on haptoglobin and haemopexin in the plasma of cattle. Australian J. Exp. Biol. Med. Sci. 42: 643-656.

Brinkmann, V., Reichard, U., Goosmann, C., Fauler, B., Uhlemann, Y., Weiss, D.S., Weinrauch, Y. and Zychlinsky, A. (2004). Neutrophil extracellular traps kill bacteria. Science 303: 1532-1535.

Brolund, L. (1985). Cell counts in bovine milk: causes of variation and applicability for diagnosis of subclinical mastitis. Acta Vet. Scand. 80: 118-123.

Brookbanks, E.O. (1966). The correlation between California mastitis test results and the presence of mastitis pathogens in composite milk samples. N Z Vet J 14:89–91. 288

Bruckmaier, R.M., Ohtsouka, E. and Blum, C.E. (2004). Fractionized milk composition in dairy cows with subclinical mastitis. Veterinary Medicine- Czech 8: 283–290.

Burvenich, C., Guidry, A.J. and Paape, M.J. (1995). Natural defence mechanisms of the lactating and dry mammary gland (overview paper). Proc. 3rd IDF Int. Mastitis Seminar, Tel Aviv, Israel.

Cairoli, F., Battocchio, M., Veronesi, M.C., Brambilla, D., Conserva, F., Eberini, I., Wait, R. and Gianazza, E. (2006). Serum protein pattern during cow pregnancy: Acute phase proteins increase in the peripartum period. Electrophoresis 27: 1617-1625.

Calder, P.C. (2008). Polyunsaturated fatty acids, inflammatory processes and inflammatory bowel diseases. Molecular Nutrition and Food Research 52: 885-897.

Camerini, S., Polici, M.L., Liotta, L.A., Petricoin, E.F. and Zhou, W. (2007). A method for the selective isolation and enrichment of carrier protein-bound low-molecular weight proteins and peptides in the blood. Proteomics Clin. Appl. 1: 176-184.

Carroll, E.J. and Schalm, O.W. (1962). Effect of deoxyribonuclease on the California Test for Mastitis. J. Dairy Sci. 45:1094.

Casado, B., Affolter, M. and Kussmann, M. (2009). OMICS-rooted studies if milk proteins, oligosaccharides and lipids. Journal of Proteomics 73: 196-208.

Caufield, W. J. and Riddell, W.H. (1935). The chloride content of cows' milk. Cornell Vet. 25:333.

Cavaletto, M., Giuffrida, M.G. and Conti, A. (2004). The proteomic approach to analysis of human milk fat globule membrane. Clinica Chimica Acta 347: 41-48.

Cavaletto, M., Giuffrida, M.G., Fortunato, D., Gardano, L., Dellavalle, G., Napolitano, L., Giunta, C., Bertino, E., Fabris, C. and Conti, A. (2002). A proteomic approach to evaluate the butyrophilin gene family expression in human milk fat globule membrane. Proteomics 2: 850–856.

Cavaletto, M., Giuffrida, M.G., Giunta, C., Vellano, C., Fabris, C., Bertino, E., Godovac- Zimmermann, and Conti A. (1999). Multiple forms of lactadherin (breast antigen BA46) and butyrophilin are secreted into human milk as major components of milk fat globule membrane. Journal of Dairy Research 66: 295-301.

Cecil, H.C., Bitman, J. and Wood, J.R. (1965). Changes in milk glycogen during mastitis. Journal of Dairy Science 48: 1607-1611.

Cerda, S. and Oh, S.K. (1990). Methods to quantitate human haptoglobin by complexation with haemoglobin. Journal of Immunological Methods 134: 51-59. 289

Chagunda, M.G.G., Larsen, T., Bjerring, M. and Ingvartsen, K.L. (2006). L-lactate dehydrogenase and N-acetyl-β-D-glucosaminidase activities in bovine milk as indicators of non-specific mastitis. Journal of Dairy Research 73: 431-440.

Chandler, R.L., Reid, I.M., Harbison, R. & France, B.R. (1974). Journal of Comparative Pathology 84:517.

Chaneton, L., Tirante, L., Maito, J., Chaves, J. and Bussman, L.E. (2008). Relationship between milk latoferrin and etiological agent in the mastitic bovine mammary gland. Journal of Dairy Science 91: 1865-1873.

Charlwood, J., Hanrahan, S., Tyldesley, R., Langridge, J., Dwek, M. and Camilleri, P. (2002). Use of proteomic methodology for the characterization of human milk fat globular membrane proteins. Analytical Biochemistry 301: 314-324.

Chertov, O., Biragyn, A., Kwak, L.W. Simpson, J.T., Boroninan, T., Hoang, V.M., Prieto, D.A., Conrads, T.P., Veenstra, T.D. and Fisher, R.J. (2004). Organic solvent extraction of proteins and peptides from serum as an effective sample preparation for detection and identification of biomarkers by mass spectrometry. Proteomics 4: 1195-1203.

Chertov, O., Simpson, J.T., Biragyn, A., Conrads, T.P., Veenstra, T.D. and Fisher, R.J. (2005). Enrichment of low-molecular-weight proteins from biofluids for biomarker discovery. Expert Rev Proteomics 2:139–145.

Cho, K., Torres, N.L., Subramanyam, S., Deepak, S.A., Sardesai, N., Han, O., Williams, C.E., Ishii, H., Iwahashi, H. and Rakwal, R. (2006). Protein extraction/solubilisation protocol for monocot and dicot plant gel-based proteomics. Journal of Plant Biology 49: 413-420.

Choe, L., D'Ascenzo, M., Relkin, N.R., Pappin, D., Ross, P., Williamson, B., Guertin, S., Pribil, P. and Lee, K.H. (2007). 8-plex quantitation of changes in cerebrospinal fluid protein expressionn in subjects undergoing intravenous immunoglobulin treatment for Alzheimer's disease. Proteomics 7:3651–3660.

Choudhary, J.S., Blackstock, W.P., Creasy, D.M. and Cottrell, J.S. (2001). Review: Matching peptide mass spectra to EST and genomic DNA database. Trends in Biotechnology 19: 17-22.

Chun, J., Yin, Y.I., Yang, G., Tarassishin, L. and Li, Y-M. (2004). Stereoselective synthesis of photoreactive peptidomimetic γ-secretase inhibitors. J. Org. Chem. 69: 7344-7347.

Cifuentes, A. (2006). Recent advances in the application of capillary electromigration methods for food analysis. Electrophoresis 27: 283-303. 290

Clynen, E., Baggerman, G., Husson, S.J., Landuyt, B., and Schoofs, L. (2008). Peptidomics in drug research. Expert Opinion on Drug Discovery. 3: 425-440.

Clynen, E., Baggerman, G., Veelaert, D., Cerstiaens, A., Van der Horst, D., Harthoorn, L., Derua, R., Waelkens, E., De Loof, A. and Schoofs, L. (2001). Peptidomics of the pars intercerebralis-corpus cardiacum complex of the migratory locust, Locusta migratoria. Europena Journal of Biochemistry 268: 1929-1939.

Clynen, E., De Loof, A. and Schoofs, L. (2003). The use of peptidomics in endocrine research. General and Comparative Endocrinology 132: 1-9.

Cohen, J.W. (1988) Statistical power analysis for the behavioural sciences (2d edn). Hillsdale, N.J. Lawrence Erlbaum Associates.

Colak, A., Polat, B., Okumus, Z., Kaya, M., Yanmaz, L.E. and Hayirli, A. (2008). Short Communication: Early detection of mastitis using infrared thermography in dairy cows. J. Dairy Sci. 91:4244-4248.

Conner, J. G., Eckersall, P. D., Wiseman, A., Aitchison, T. C.and Douglas, T. A. (1988). Bovine acute phase proteins after turpentine injection. Res. Vet. Sci. 44: 82-88.

Conner, J. G., Eckersall, P. D., Wiseman, A., Bain, R. K. & Douglas, T. A. (1989). Acute phase response in calves following infection with Pasteurella haemolytica, Ostertagia ostertagi and endotoxin administration. Research in Veterinary Science 47: 203-207.

Conner, J.G., Eckersall, P.D., Doherty, M. and Douglas, T.A. (1986). Acute phase response and mastitis in the cow. Research in Veterinary Science 41: 126–128.

Connolly, J.P., Comerci, D., Alefantis, T.G., Walz, A., Quan, M., Chafin, R., Grewal, P., Mujer, C.V., Ugalde, R.A. and DelVecchio, V.G. (2006). Proteomic analysis of Brucella abortus cell envelope and identification of immunogenic candidate proteins for vaccine development. Proteomics 6: 3767-3780.

Cooray, R. (1994). Use of bovine myeloperoxidase as an indicator of mastitis in dairy cattle. Veterinary Microbiology 42 317-326.

Cooray, R., Petersson, C.B.G. and Holmberg, O. (1993). Isolation and purification of bovine myeloperoxidase from neutrophil granules. Vet. lmmunol. Immunopathol. 38:261-272.

Corray, R., Waller, K.P. and Venge, P. (2007). Haptoglobin comprises about 10% of granule protein extracted from bovine granulocytes isolated from healthy cattle. Veterinary Immunology and Immunopathology 119: 310-315. 291

Couté, Y., Hernandez, C., Appel, R. D., Sanchez, J. C. & Margolles, A. (2007). Labeling of Bifidobacterium longum cells with 13C-substituted leucine for quantitative proteomic analyses. Appl Environ Microbiol 73: 5653–5656.

Cravatt, B.F., Simon, G.M. & Yates, J.R. (2007). The biological impact of mass-spectrometry-based proteomics. Nature 450: 991-1000.

Creamer, L.K. and Olson, N.F. (1982). Rheological evaluation of maturing cheddar cheese. Journal of Food Science 47: 631-636.

Crossman, J.V., Dodd, F.H., Lee, J.M. and Neave, F.K. (1950). The effect of bacterial infection on the milk yield of the individual quarters or the cow’s udder. J. Dairy Res. 17:128–158.

D’Amato, A., Bachi, A., Fasoli, E., Boschetti, E., Peltre, G., Sénéchal, H. and Righetti, P.G. (2009). In-depth exploration of cow’s whey proteome via combinatorial peptide ligand libraries. Journal of Proteome Research 8: 3925-3936.

Daniel, R.C.W., Biggs, D.A. and Barnum D.A (1966). The relationship between California Mastitis Test scored and monthly milk production and composition. Can. Vet. Jour. 7: 99-105.

Danielsen, M., Codrea, M.C., Ingvartsen, K.L., Friggens, N.C., Bendixen, E. and Rontved, C.M. (2010). Quantitative milk proteomics-Host responses to lipopolysaccharide-mediated inflammation of bovine mammary gland. Proteomics 10: 2240-2249.

Davis, J.G. (1947). The rapid abnormality indicator. Dairy Industries 12:35-41.

Davis, M.T., Beierle, J., Bures, E.T., McGinley, M.D., Mort, J., Robinson, J.H., Spahr, C.S., Yu, W., Luethy, R. and Patterson, S.D. (2001). Automated LC-LC-MS-MS platform using binary ion- exchange and gradient reversed-phase chromatography for improved proteomic analyses. Journal of Chromatography B: Biomedical Sciences and Applications 752: 281-291.

De Frutos, M., Cifuentes, A., Diez-Masa, J.C., Amigo, L. and Ramos, M. (1991). Application of HPLC for the detection of proteins in whey mixtures from different animal species. HRC Journal of High Resolution Chromatography 14: 289-291.

De Mol, R.M. and Ouweltjes, W. (2001). Detection model for mastitis in cows milked in an automatic milking system. Prev. Vet. Med. 49: 71-82.

Defra, Agriculture in the United Kingdom 2009, April 2010, Table 9.1: http://www.defra.gov.uk/evidence/statistics/foodfarm/general/auk/latest/excel/index.htm 2008/09/080917145736.html. 292

Derzelle, S., Bolotin, A., Mistou, M.Y. and Rul, F. (2005). Proteome analysis of Streptococcus thermophilus grown in milk reveals pyruvate formate-lyase as the major upregulated protein. Appl. Environ. Microbiol. 71: 8597–8605.

Detilleux, J.C., Kehrli, M.E., Freeman, A.E., Fox L.K. and Kelley, D.H. (1995). Mastitis of periparturient Holstein cattle: a phenotypic and genetic study. Journal of Dairy Science 78: 2285- 2293.

Dettmer, K., Aronov, P.A. and Hammock, B.D. (2007). Mass spectrometry-based metabolomics. Mass Spectrom. Rev. 26: 51-78.

Dhiman, T.R., Anand, G.R., Satter, L.D. and Pariza, M.W. (1999). Conjugated linoleic acid content of milk from cows fed different diets. Journal of Dairy Science 82: 2146-2156.

Dibartola, S.P. and Benson, M.D. (1989). The pathogenesis of reactive systemic amyloidosis. Journal of Veterinary Internal Medicine 3: 31-41.

Dijkman, A.J. (1975). Cells in milk. Neth. Milk and Dairy J. 29: 99-103.

Dingwell, R.T., Leslie, K.E., Sabour, P., Lepp, D. and Pacan, J. (2006). Influence of the genotype of Staphylococcus aureus, determined by pulsed-field gel electrophoresis, on dry-period elimination of subclinical mastitis in Canadian dairy herds. Canadian Journal of Veterinary Research-Revue 70: 115-120.

Dingwell, R.T., Leslie, K.E., Schukken, Y.H., Sargeant, J.M. & Timms, L.L. (2003). Evaluation of the California Mastitis Test to detect an intramammary infection with a major pathogen in early lactation dairy cows. Can Vet J 44:413–416.

Doherty, M.K., Beynon, R.J. and Whitfield, P.D. (2008). Proteomics and naturally occurring animal diseases: Opportunities for animal and human medicine. Proteomics Clinical Applications 2: 135-141.

Dohoo, I.R. and Meekk, A.H. (1982). Somatic cell counts in bovine milk. Can Vet J. 23:119-125.

Douglas, G. N., Overton, T.R., Bateman, II H.G., Dann, H.M. and Drackley, J.K. (2006). Prepartal plane of nutrition, regardless of dietary energy source, affects periparturient metabolism and dry matter intake in Holstein cows. J. Dairy Sci. 89: 2141-2157.

Dowell, J.A., Frost, D.C., Zhang, J. and Li, L. (2008). Comparison of two-dimensional fractionation techniques for shotgun proteomics. Anal. Chem. 80: 6715-6723.

Dunn, W.B. and Ellis, D.I. (2005). Metabolomics: current analytical platforms and methodologies. Trends Anal. Chem. 24: 285–294. 293

Eckersall, P.D. (1995). Acute phase proteins as markers of inflammatory lesions. Comparative Haematology International 5: 93-97.

Eckersall, P.D. and Conner, J.G. (1988). Bovine and canine acute phase proteins. Veterinary Research Communications 12: 169-178.

Eckersall, P.D. and Conner, J.G. (1990). Plasma haptoglobin in cattle (Bos Taurus) exists as polymers in association with albumin. Comp. Biochem. Physiol 96B:309-314.

Eckersall, P.D., Parton, H., Conner, J.G., Nash, A.S., Watson, T. and Douglas, T.A. (1988). Acute phase reactants in diseases of dog and cattle. In: Blackmore, D.J., Eckersall, P.D., Evans, G.O. et al. (eds) Animal clinical biochemistry: the future. Cambridge; Cambridge University Press, 225–230.

Eckersall, P.D., Young, F.J., McComb, C., Hogarth, C.J., Safi, S., Weber, A., McDonald, T., Nolan, A.M. and Fitzpatrick, J.L. (2001). Acute phase proteins in serum and milk from dairy cows with clinical mastitis. Veterinary Record 148: 35-41.

Eckersall, P.D., Young, F.J., Nolan, A.M., Knight, C.H., McComb, C., Waterston, M.M., Hogarth, C.J., Scott, E.M. and Fitzpatrick, J.L. (2006). Acute phase proteins in bovine milk in an experimental model of Staphylococcus aureus subclinical mastitis. Journal of Dairy Science 89: 1488-1501.

Eddy, A. L., van Hoogmoed, L.M. and Snyder, J.R. (2001). Review: The role of thermography in the management of equine lameness. Vet. J. 162:172–181.

Eicher, S. D., Cheng, H.W., Sorrells, A.D. and Schutz, M.M. (2006). Behavioral and physiological indicators of sensitivity or chronic pain following tail docking. J. Dairy Sci. 89:3047-3051.

Eigel, W.N., Butler, J.E., Ernstrom, C.A., Farrell, Jr. H.M., Harwalkar, V.R., Jenness, R. and Whitney, R.M. (1984). Nomenclature of proteins of cow’s milk: 5th revision. Journal of Dairy Science 67: 1599-1631.

El Ghmati, S.M., Van Hoeyveld, E.M., Van Strijp, J.G., Ceuppens, J.L. and Stevens, E.A. (1996). Identification of haptoglobin as an alternative ligand for CD11b/CD18. J Immunol 156: 2542–2552.

Elson, E.C. (1974). Quantitative determination of serum haptoglobin. A simple and rapid method. American Journal of Clinical Pathology 62: 655-663.

Erskine, R.J. and Bartlett, P.C. (1993). Serum concentrations of copper, iron and zinc during Escherichia coli-induced mastitis. Journal of Dairy Science 76: 408-413.

Etzel, K.R., Swerdel, M.R., Swerdel, J.N. and Cousins, R.J. (1982). Endotoxin-induced changes in copper and zinc metabolism in the Syrian hamster. J. Nutr. 112: 363-2373. 294

Evans, E.W. (1982). Uses of milk proteins in formulated foods. In B. J. F. Hudson (Ed.), Developments in Food Proteins. London: Applied Science.

Fenn, J. B., Mann, M., Meng, C.K., Wong, S.F. and Whitehouse, C.M., (1989). Electrospray ionization for mass spectrometry of large biomolecules. Science 246:64–71.

Fernando, R.S., Rindsig, R.B. and Spahr, S.L. (1982). Electrical conductivity of milk for detection of mastitis. J. Dairy Sci 65:659-664.

Fernando, R.S., Spahr, S.L. and Jaster, E.H. (1985). Comparison of electrical conductivity of milk with other indirect methods for detection of subclinical mastitis. J. Dairy Sci 68:449-456.

Ferrand, M., Huquet, B., Barbey, S., Barillet, F., Faucon, F., Larroque, H., Leray, O., Trommenschlager, J.M. and Brochard, M. (2011). Determination of fatty acid profile in cow’s milk using mid-infrared spectrometry: Interest of applying a variable selection by genetic algorithms before a PLS regression. Chemometrics and Intelligent Laboratory Systems 106: 183-189.

Feuerstein, G. and Hallenbeck, J.M. (1987). Prostaglandins, leukotrienes and platelet-activating factor in shock. Ann. Rev. Pharmacol. Toxicol. 27: 301-313.

Fiedler, G.M., Baumann, S., Leichtle, A., Oltmann, A., Kase, J., Thiery, J. and Ceglarek, U. (2007). Standardized peptidome profiling of human urine by magnetic bead separation and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Clin Chem. 53:421–428.

FitzGerald, C.H., Deeth, H.C. & Kitchen, B.J. (1981). The relationship between the levels of free fatty acids, lipoprotein lipase, carboxylesterase, N-acetyl-p-D-glucosaminidase, somatic cell count and other mastitis indices in bovine milk Journal of Dairy Research 48:253-265.

Fong, B.Y. and Norris, C.S. (2009). Quantification of milk fat globule membrane proteins using selected reaction monitoring mass spectrometry. J. Agri. Food Chem. 57: 6021-6028.

Fong, B.Y., Norris, C.S. and Palmano, K.P. (2008). Fractionation of bovine whey proteins and characterisation by proteomic techniques. International Dairy Journal 18: 23-46.

Forbes, D. & Hebert, C.N. (1968). Studies in the pathogenesis of staphylococcal mastitis. Veterinary Record 82:69-73.

Forster, T.L., Ashworth, U.S. and Luedecke, L.O., (1966). Relationship between California Mastitis Test reaction and production and composition of milk from opposite quarters. J. Dairy Sci 50:675- 682. 295

Fox, P.F. and Brodkorb, A. (2008). The casein micelle: Historical aspects, current concepts and significance. International Dairy Journal 18: 677-684.

Fox, P.F. and Kelly, A.L., (2003). Developments in the chemistry and technology of milk proteins. 2. Minor milk proteins. Food Australia 55: 231-234.

Gagnaire, V., Jardin, J., Jan, G. and Lortal, S. (2009). Invited review: Proteomics of milk and bacteria used in fermented dairy products: From qualitative to quantitative advances. Journal of Dairy Sciences 92: 811-825.

Galvani, M., Hamdan, M. and Righetti, P.G. (2000). Two-dimensional gel electrophoresis/matrix- assisted laser desorption/ionisation mass spectrometry of a milk powder. Rapid Communications in a Mass Spectrometry 14: 1889-1897.

Galvani, M., Hamdan, M. and Righetti, P.G. (2001). Two-dimensional gel electrophoresis/matrix- assisted laser desorption/ionisation mass spectrometry of commercial bovine milk. Rapid Communications in Mass Spectrometry 15: 258-264.

Ganheim, C., Hulten, C., Carlsson, U., Kindahl, H., Niskanen, R. and Waller, K.P. (2003). The acute phase response in calves experimentally infected with bovine viral diarrhea virus and/or Mannheimia heamolytica. J Vet. Med. B. 50: 183-190.

Gerardi, G., Bernardini, D., Elia, C.A., Ferrari, V., Iob, L. and Segato, S. (2009). Use of serum amyloid A and milk amyloid A in the diagnosis of subclinical mastitis in dairy cows. J. Dairy Res. 76: 411-417.

Giesecke, W. H. & Van Den Heever, L.W. (1981). Levels of glucose, serum albumin and somatic cells before and during early stages of acute clinical mastitis artificially induced in cows by means of human strains of group B streptococci (GBS) administered intracisternally. Onderstepoort Journal of Veterinary Research 48: 69-75.

Giesecke, W.H. (1975). The definition of bovine mastitis and the diagnosis of its subclinical types during normal lactation. International Dairy Federation Annual Bulletin, Document 85:62-70.

Giesecke, W.H. & Viljoen, M.H. (1974). The diagnosis of subclinical mastitis in lactating cows: a comparison of cytologieal methods and a monovalent radial immunodiffusion test. Onderstepoort Journal of Veterinary Research 41:51-74.

Giri, S. N., Chen, Z., Carroll, E.J., Mueller, R., Schiedt, M.J. and Panico, L. (1984). Role of prostaglandins in pathogenesis of bovine mastitis induced by Escherichia coli endotoxin. Am. J. Vet. Res. 45:586–591. 296

Gitton, C., Meyrand, M., Wang, J., Caron, C., Trubuil, A., Guillot, A. and Mistou, M.Y. (2005). Proteomic signature of Lactococcus lactis NCDO763 cultivated in milk. Appl. Environ. Microbiol. 71: 7152–7163.

Godson, D.L., Campos, M., Attah-Poku, S.K., Redmond, M.J., Cordeiro, D.M., Sethi, M.S., Harland, R.J. and Babiuk, L.A. (1996). Serum haptoglobin as an indicator of the acute phase response in bovine respiratory disease. Veterinary Immunology and Immunopathology 51: 277-292.

Goff, J.P. (2006). Major advances in our understanding of nutritional influences on bovine health. J. Dairy Sci. 89: 1292-1301.

Gordon, A.H. and Koy, A. (1985). The Acute Phase Response to Injury and Infection. The Roles of Interleukin 1 and Other Mediators. Amsterdam: Elsevier, ISBN: 0444 80648-2.

Görg, A., Boguth, G., Obermaier, C., Hader, A. and Weiss, W. (1998). 2-D electrophoresis with immobilized pH gradients using IPGphor Isoelectric Focusing System. Life Science News 1: 1-3.

Goswami, S.N., Roy, A. & Kalyani, I.H. (2003). A comparative study on various indirect tests to direct cultural isolation for detection of subclinical mastitis (SCM). In: Proceedings of XXI Indian Society for Veterinary Medicine (ISVM) Conference, 13–15 February 2003 Anand, India. pp. 101– 102.

Graham, D.R.M., Elliott, S.T. and Van Eyk, J.E. (2005). Broad based proteomic strategies: A practical guide to proteomic functional screening. The Journal of Physiology 563: 1-9.

Grandison, A.S. and Ford, G.D. (1986). Effects of variations in somatic cell count on the rennet coagulation properties of milk and on the yield, composition and quality of Cheddar cheese. J. Dairy Res. 53: 645–655.

Grant, I.R. (2005). Zoonotic potential of Mycobacterium avium ssp paratuberculosis: the current position. Journal of Applied Microbiology 98: 1282-1293.

Greatrix, G. R., Quayle, J. C. & Coombe, R. A. (1968). The detection of abnormal milk by electrical means. Journal of Dairy Research 35: 213-221.

Greenbaum, D., Medzihradszky, K.F., Burlingame, A. and Bogyo, M. (2000). Epoxide electrophiles as activity-dependent cysteine protease profiling and discovery tools. Chemistry & Biology 7: 569- 581. 297

Greenbaum, D.C., Baruch, A., Grainger, M., Bozdech, Z., Medzihradszky, K.F., Engel, J., Derisi, J., Holder, A.A. & Bogyo, M. (2002). A role for the protease falcipain 1 in host cell invasion by the human malaria parasite. Science 298: 2002-2006.

Griinari, J.M., Dwyer, D.A., McGuire, M.A., Bauman, D.E., Palmquist, D.L. & Nurmela, K.V.V. (1998). Trans-Octadecenoic acids and milk fat depression in lactating dairy cows. Journal of Dairy Science 81: 1251-1261.

Grönlund, U., Hultén, C., Eckersall, P.D., Hogarth, C. and Waller, K.P. (2003). Haptoglobin and serum amyloid A in milk and serum during acute and chronic experimentally induced Staphylococcus aureus mastitis. Journal of Dairy Research 70: 379-386.

Grönlund, U., Sandgren, C.H. and Waller, K.P. (2005). Haptoglobin and serum amyloid A in milk from dairy cows with chronic sub-clinical mastitis. Vet. Res. 36: 191-198.

Gruys, E., Obwolo, M.J. and Toussaint, M.J.M. (1994). Diagnostic significance of the major acute phase proteins in veterinary clinical chemistry: a review. Veterinary Bulletin 64: 1009–1018.

Gruys, E., Toussaint, M.J.M., Landman, W.J.M., Tivapasi, M., Chamanza, R. & van Veen, L. (1999) Infection, Inflammation and Stress Inhibit Growth. Mechanisms and Non-specific Assessment of the Processes by Acute Phase Proteins. In: Wensing, T. (Ed.), Production Diseases in Farm Animals. 10th International Conference, 1998. Wageningen Press: Wageningen, 72-87. ISBN: 90-74134-60-2.

Gruys, E., Toussaint, M.J.M., Niewold, T.A. and Koopmans, S.J. (2005). Acute phase reaction and acute phase proteins. Journal of Zhejiang University SCIENCE 6B: 1045-1056.

Guarino, C., Fuselli, F., La, M.A., Longo, L., Faberi, A. and Marianella, R.M. (2010). Peptidomic approach, based on liquid chromatography/electrospray ionization tandem mass spectrometry, for detecting sheep’s milk in goat’s and cow’s cheeses. Rapid Communications in Mass Spectrometry 24: 705–713.

Gundry, R.L., Fu, Q., Jelinek, C.A., Van, Eyk, J.E., Cotter, Dr, R.J. (2007). Investigation of an albumin-enriched fraction of human serum albumin and its albumimome. Proteomics-Clinical Applications 1: 73-88.

Gygi, S.P., Rist, B., Gerber, S.A., Turecek, F., Gelb, M.H. and Aebersold, R. (1999). Quantitative analysis of complex protein mixtures using isotope-coded affinity tags. Nat Biotechnol. 17: 994-999.

Haenlein, G. T. W., Schultz, L.H. and Zikakis, J.P. (1973). Comparison of proteins in milk with varying leukocyte contents. J. Dairy Sci., 56: 1017-1023. 298

Haghkhah, M., Nazifi, S. and Jahromi, A.G. (2010). Evaluation of milk haptoglobin and amyloid A in high producing dairy cattle with clinical and subclinical mastitis in Shiraz. Comparative Clinical Pathology 19: 547-552.

Halasa, T., Huijps, K., Osteras, O. and Hogeveen, H. (2007). Economic effects of bovine mastitis and mastitis management: A review. Veterinary Quarterly 29: 18-31.

Hale, J.E., Gelfanova, V., Ludwig, J.R. and Knieman, M.D. (2003). Application of proteomics for discovery of protein biomarkers. Briefings in Functional Genomics and Proteomics 2: 185-193.

Hamann, J. and Kromker, V. (1997). Potential of specific milk composition variables for cow health management. Livestock Prod. Sci. 48: 201-208.

Hamed, H., El Feki, A. and Gargouri, A. (2008). Total and differential bulk cow milk somatic cell counts and their relation with antioxidant factors. Comptes Rendus Biologies 331: 144-151.

Harmon, R. J. (1994). Physiology of mastitis and factors affecting somatic cell counts. J. Dairy Sci., 77:2103-2112.

Harvey, J.W. (1976). Quantitative determinations of normal horse, cat and dog haptoglobins. Theriogenoloy 6: 133-151.

He, Q.Y., and Chiu, J.F. (2003). Proteomics in biomarker discovery and drug development. Journal of Cellular Biochemistry 89: 868-886.

Head, M. J. and Dyson, S. (2001). Talking the temperature of equine thermography. Vet. J. 162:166– 167.

Heegaard, P.M.H., Godson, D.L., Toussaint, M.J.M., Tjørnehøj, K., Larsen, L.E., Viuff, B. and Rønsholt, L. (2000). The acute phase response of haptoglobin and serum amyloid A (SAA) in cattle undergoing experimental infection with bovine respiratory syncytial virus. Veterinary Immunology and Immunopathology 77: 151-159.

Hennig-Pauka, I., Jacobsen, I., Blecha, F., Waldmann, K.H. and Gerlach, G.F. (2006). Differential proteomic analysis reveals increased cathelicidin expression in porcine bronchoalveolar lavage fluid after an Actinobacillus pleuropneumoniae infection. Veterinary Research 37: 75-87.

Herdt, T.H. (2000). Ruminant adaptation to negative energy balance. Influences on the etiology of ketosis and fatty liver. Veterinary Clinics of North America Food Animal Practice 16: 215–230. 299

Hettinga, K.A., van Valenberg, H.J.D., Lam, T.J.G.M., van Hooijdonk, A.C.M. (2009). The influence of incubation on the formation of volatile bacterial metabolites in mastitis milk. Journal of Dairy Science 92: 4901-4905.

Hettinga, K.A., van Valenberg, H.J.F., Lam, T.J.G.M. and van Hooijdonk, A.C.M. (2008). Detection of mastitis pathogens by analysis of volatile bacterial metabolites. Journal of Dairy Science 91: 3834- 3639.

Hettinga, K.A., van Valenberg, H.J.F., Lam, T.J.G.M. and van Hooijdonk, A.C.M. (2009). The origin of the volatile metabolites found in mastitis milk. Veterinary Microbiology 137: 384-387.

Hiendleder, S., Bauersachs, S., Boulesteix, A., Blum, H., Arnold, G.J., Frohlich, T. and Wolf, E. (2005). Functional genomics: tools for improving farm animal health and welfare. Revue Scientifique et Technicue 24:355-377.

Hillerton, J.E. (1999a). Balancing mastitis and quality. Proceedings of the British Mastitis Conference 31-36.

Hillerton, J.E. (1999). Redefining mastitis based on somatic cell count. International Dairy Federation 345: 4-6.

Hillerton, J.E. (2000). Detecting mastitis cow-side. National Mastitis Council Annual Meeting Proceedings 48-53.

Hirvonen, J., Hietakorpi, S. & Saloniemi, H. (1997). Acute phase response in emergency slaughtered dairy cows. Meat Science 46: 249-257.

Hirvonen, J., Pyorala, S. and Jousimies-Somer, H. (1996). Acute phase response in heifers with experimentally induced mastitis. Journal of Dairy Research 63: 351-360.

Hiss, S., Mielenz, M., Bruckmaier, R.M. and Sauerwein, H. (2004). Haptoglobin concentrations in blood and milk after endotoxin challeng and quantification of mammary Hp mRNA expression. Journal of Dairy Science 87: 3778-3784.

Hogan, J.S., Gonzalez, R.N., Harmon, R.J. et al. (1999). Sample collection and handling; testing procedure. In: Hogan JS, Gonzalez RN, Harmon RJ, et al, eds. Laboratory handbook on bovine mastitis. Revised edition. Madison, Wis: National Mastitis Council Inc,; 1–11, 205–212.

Hogan, J.S., Pankey, J.W. and Duthie, A.H. (1987). Growth inhibition of mastitis pathogens by long- chain fatty acids. Journal of Dairy Science 70: 927-934. 300

Hogarth, C.J., Fitzpatrick, J.L., Nolan, A.M., Harris, N. and Eckersall, P.D. (2002). Proteomics to study changes in bovine milk during mastitis. Research in Veterinary Science 72: 23.

Hogarth, C.J., Fitzpatrick, J.L., Nolan, A.M., Young, F.J., Pitt, A. and Eckersall, P.D. (2004.) Differential protein composition of bovine whey: A comparison of whey from healthy animals and from those with clinical mastitis. Proteomics 4: 2094-2100.

Hogeveen, H., Miltenburg, J.D., den Hollander, S. and Drandena, K. (2001). Milking three times a day and its effect on udder health and production. Intl. Dairy Fed. Mastitis Newsletter 24: 7.

Holdaway, R.J., Holmes, C.W. and Steffert, I.J. (1996). A comparison of indirect methods for diagnosis of subclinical intramammary infection in lactating dairy cows. 2. The discriminative ability of eight parameters in foremilk from individual quarters and cows. Aust. J. Dairy Technol. 51: 72-78.

Holland, J.W., Deeth, H.C. & Alewood, P.F. (2004). Proteomic analysis of κ-casein micro- heterogeneity. Proteomics 4: 743-752.

Holland, J.W., Deeth, H.C. and Alewood, P.F. (2006). Resolution and characterisation of multiple isoforms of bovine κ-casein by 2-DE following a reversible cysteine-tagging enrichment strategy. Protoemics 6: 3087-3095.

Hollis, B.W., Roos, B.A., Draper, H.H. and Lambert, P.W. (1981). Vitamin D and its metabolites in human and bovine milk. Journal of Nutrition 111: 1240-1248.

Honkanen-Buzalski, T. and Sandholm, M. (1981). Trypsin-inhibitora in mastitic milk and colostrum: correlation between trypsin-inhibitor capacity, bovine serum albumin and somatic cell contents. Journal of Dairy Research 48: 213-223.

Honkanen-Buzalski, T., Katilla, T. and Sandholm, M. (1981). Milk antitrypsin activity during clinical and experimental bovine mastitis. Acta Veterinaria Scandinavica 22: 360-368.

Hopster, H., van der Werf, J.T.N. and Blokhuis, H.J. (1998). Stress enhanced reduction in peripheral blood lymphocyte numbers in dairy cows during endotoxin-induced mastitis. Veterinary Immunology and Immunopathology 66: 83-97.

Horadagoda, A., Eckersall, P.D. and Alsemgeest, S.P.M. (1993). Purification and quantitative measurement of bovine serum amyloid-A. Research in Veterinary Science 55: 317-325.

Horadagoda, N.U., Knox, K.M.G., Gibbs, H.A., Rease, S.W.J., Horadagoda, A., Edwards, S.E.R. and Eckersall, P.D. (1999). Acute phase proteins in cattle: discrimination between acute and chronic inflammation. Vet. Rec. 144: 437–441. 301

Hovinen, M., Siivonen, J., Taponen, S., Hänninen, L., Pastell, M., Aisla, A.-M. and Pyörälä, S. (2008). Detection of Clinical Mastitis with the Help of a Thermal Camera J. Dairy Sci. 91:4592– 4598.

Hu, L.H., Ye, M.L. and Zou, H.F. (2009). Recent advances in mass spectrometry-based peptidome analysis. Expert Review of Proteomics 6: 433-447.

Hughes, V., Smith, S., Garcia-Sanchez, A., Sales, J. and Stevenson, K. (2007). Proteomic comparison of Mycobacterium avium subspecies paratuberculosis grown in vitro and isolated from clinical cases of ovine paratuberculosis. Microbiology-SGM 153: 196-205.

Hunt, D.F., Henderson, R.A., Shabanowitz, J., Sakaguchi, K., Michel, H., Sevilir, N., Cox, A.L., Appella, E. and Engelhard, V.H. (1992). Characterization of peptides bound to the class IMHC molecule HLA-A2.1 by mass spectrometry. Science 255:1261–1263.

Hurnik, J. F., DeBoer, S. and Webster, A.B., (1984). Detection of health disorders in dairy cattle utilizing a thermal infrared scanning technique. Can. J. Anim. Sci. 64:1071–1073.

Husebekk, A., Husby, G., Sletten, K., Skogen, B. and Nordstoga, K. (1988). Characterization of bovine amyloid proteins SAA and AA. Scand J Immunol 27: 739-743.

Husebekk, A., Skogen, B. and Husby, G. (1987). Characterization of amyloid protein-AA and protein-SAA as apolipoproteins of high-density-lipoprotein (HDL)-Displacement of SAA from the HDL-SAA complex by APO AI and APO AII. Scandinavian Journal of Immunology 25: 375-381.

Husebekk, A., Skogen, B., Husby, G. and Marhaug, G. (1985). Transformation of amyloid precursor- SAA to protein-AA and incorporation in amyloid fibrils in vivo. Scandinavian Journal of Immunology 21: 283-287.

Huszenicza, G., Janosi, S., Kulcsar, M., Korodi, P., Dieleman, S.J. Bartyik, J., Rudas, P. and Ribiczei- Szabo, P. (1998). Gram-negative mastitis in early lactation may interfere with ovarian and certain endocrine functions and metabolism in dairy cows. Reproduction in Domestic Animals 33: 147-153.

Hutchens, T.W. and Yip, T.T. (1993). New desorption strategies for the mass spectrometric analysis of macromolecules. Rapid Commun Mass Spectrom 7: 576-580.

Ibeagha-Awemu, E.M., Ibeagha, A.E., Messier, S. and Zhao X. (2010). Proteomics, genomics and pathway analyses of Escherichia coli and Staphylococcus aureus infected milk whey reveal molecular pathways and networks involved in mastitis. Journal of Proteome Research 9: 4604-4619. 302

Illyin, S.E., Belkowski, S.M. and Plata-Salaman, C.R. (2004). Biomarker discovery and validation: technologies and integrative approaches. Trends in Biotechnology 22: 411-416.

Ingvartsen, K.L. and Friggens, N.C. (2005). To what extent do variabilities in hormones, metabolites and energy intake explain variability in milk yield? Domest. Anim. Endocrinol. 29: 294–304.

Isaaq, H.J., Conrads, T.P., Prieto, D.A., Tirumalai, R. and Veenstra, T.D. (2003). SELDI-TOF MS for diagnostic proteomics. Analytical Chemistry 75: 148-155.

Isaaq, H.J., Veenstra, T.D., Conrads, T.P. and Felschow, D. (2002). The SELDI-TOF MS approach to proteomics: protein profiling and biomarker identification. Biochem. Biophys. Res. Commun. 292: 587–592.

Issaq, H. J. (2001). The role of separation science in proteomics research. Electrophoresis 22:3629– 3638.

Issaq, H. J., Conrads, T. P., Janini, G. M. & Veenstra, T. D. (2002). Methods for fractionation, separation and profiling of proteins and peptides. Electrophoresis 23:3048–3061.

Ivanov, A., Schimdt, M., Doud, M. and Wolf, D. (2005). Deciphering complex proteomes using liquid and gel phase prefractionation approaches. Molecular & Cellular Proteomics 4: S340.

Ivanov, V.T., Blishchenko, E.Y. and Karelin, A.A. (2009). Functional peptidomics: Recent developments and state of art. Innovations in Chemical Biology Editor: Sener B. pp. 15-27.

Jacobsen, S., Andersen, P.H., Toelboell, T. and Heegaard, P.M.H. (2004). Dose dependency and individual variability of the lipopolysaccharide-induced bovine acute phase protein response. J. Dairy Sci. 87: 3330–3339.

Jensen, L.E. and Whitehead, A.S. (1998). Regulation of serum amyloid A protein expression during the acute-phase response. Biochem. J. 15: 489-503.

Jensen, P.T. (1957). Investigations into the Whiteside Test and the CMT for the detection of pathological secretions in herd milk samples. Nord. Vet. Med. 9:590-608.

Jensen, R.C, Ferris, A. M. and Lammi-Keefe, C.J. (1991). The composition of milk fat. J. Dairy Sci. 74, 3228-3243.

Jensen, R.G. (1995). Handbook of Milk Composition. Milk Lipid 566. San Diego, CA: Academic Press. 303

Jia, X.H., Hildrum, K.I., Westad, F., Kummen, E., Aass, L. and Hollung, K. (2006b). Changes in enzymes associated with energy metabolism during early post mortem period in Longissimus thoracis bovine muscle analyzed by proteomics. Journal of Proteome Research 5: 1763-1769.

Jia, X.H., Hollung, K., Therkildsen, M., Hildrum, K.I. and Bendixen, E. (2006a). Proteome analysis of early post-mortem changes in two bovine muscle types: M-longissimus dorsi and M- semitendinosis. Proteomics 6: 936-944.

Jones, G.E. and Mould, D.L. (1984). Adaptation of the guaiacol (peroxidase) test for haptoglobins to a microtitration plate system. Research in Veterinary Science 37: 87–92.

Kaartinen L. & Jensen, N.E. (1988). Use of N-acetyl-β-D-glucosaminidase to detect teat canal inflammations. Journal of Dairy Research 55:603-607.

Karas, M. and Hillenkamp, F. (1988). Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons. Anal Chem 60: 2299.

Karelin, A.A., Philippova, M.M., Karelina, E.V., Strizhkov, B.N., Grishnina, G.A., Nazimov, I.V. and Ivanov, V.T. (1998). Journal of Peptide Science 4: 211-225.

Karijord, Ø., Standal, N. and Syrstad, O. (1982). Sources of variation in composition of milk fat. Z. Tierzüchtg. Züchtgsbiol. 99: 81-93.

Kato, K., Mori, K. & Katoh, N. (1989). Contribution of leucocytes to the origin of lactate dehydrogenase isozymes in milk of bovine mastitis. Japanese Journal of Veterinary Science 51:530– 539.

Katoh, N. (2002). Relevance of apolipoproteins in the development of fatty liver and fatty liver- related peripartum diseases in dairy cows. J. Vet. Med. Sci. 64: 293-307.

Kawakami, A., Hirayama, K., Kawakami, F., Kawakami, H. and Fujihara, M. (2006). Purification and biochemical characterization of a fibroblast growth factor-binding protein (FGF-BP) from the lactoferrin fraction of bovine milk. Biochimica et Biophysica Acta (BBA) 1760: 421-431.

Kehrli, M. E. Jr. and Shuster, D.E. (1994). Factors affecting milk somatic cells and their role in health of the bovine mammary gland. J. Dairy Sci. 77:619–627.

Kell, D.B. (2006). Systems biology, metabolic modelling and metabolomics in drug discovery and development. Drug Discovery Today 11: 1085-1092. 304

Kell, D.B. and Mendes, P. (1999) In: A. Cornish-Bowden, M.L. Cardenas (Editors), Technological and Medical Implications of Metabolic Control Analysis, Kluwer Academic Publishers, London, UK, p. 3.

Kelly, A.L. and McSweeney, P.L.H. (2003). Indigenous proteolytic enzymes in milk. In: P.F. Fox and P.L.H. McSweeney, Editors, Advanced dairy chemistry I, proteins (2nd ed.), Kluwer Academic- Plenum Publishers, New York, USA, pp. 495–544.

Kelly, A.L., O’Flaherty, F.O. and Fox, P.F. (2006). Indigenous proteolytic enzymes in milk: A brief overview of the present state of knowledge. International Dairy Journal 16: 563-572.

Kelly, V.C., Kuy, S., Palmer, D.J., Xu, Z.Z., Davis, S.R. and Cooper, G.J. (2006). Characterization of bovine seminal plasma by proteomics. Proteomics 6: 5826-5833.

Kessler, B.M., Tortorella, D., Altun, M., Kisselev, A.F., Fiebiger, E., Hekking, B.G., Ploegh, H.L. and Overkleeft, H.S. (2001). Extended peptide-based inhibitors efficiently target the proteasome and reveal overlapping specificities of the catalytic beta-subunits. Chem Biol. 8: 913–929.

Kidd, D., Liu, Y. and Cravatt, B.F. (2001). Profiling serine hydrolase activities in complex proteome. Biochemistry 40: 4005-4015.

Kiekermeier, F. and Güll, J. (1966). Uber das Vorkommender B-Glucuronidase in Kuhmilch. Naturwissenschaften 23:613-614.

Kimmel, E., Arkin, H. and Berman, A. (1992). Evaporative cooling of cattle: Transport phenomena and thermovision. Page 14 in Papers of the Am. Soc. Agric. Eng., St. Joseph, MI.

Kinsella, J.E. and Morr, C.V. (1984). Milk proteins: physicochemical and functional properties. CRC Critical Reviews in Food Sciences and Nutrition 21: 201-202.

Kirschke, H. and Barrett, A. (1987). Chemistry of lysosomal proteases. In: Glaumann H, Ballard FJ (eds) Lysosomes: their role in protein breakdown. Academic Press, London pp. 193 237.

Kisza, J. and Batura, K. (1969). Composition and properties of normal and mastitis milk fat. Milchwissenschaft 24: 465-468.

Kitchen, B. J. (1976). Enzymatic methods for estimation of the somatic cell count in bovine milk. I. Development of assay techniques and a study of their usefulness in evaluating the somatic cell count in milk. J. Dairy Res. 43:251-258.

Kitchen, B.J. (1981). Review of the progress of dairy science: Bovine mastitis: Milk compositional changes and related diagnostic tests. J. Dairy Sci. 64: 167-188. 305

Kitchen, B.J. and Middleton, G. (1976) Enzymic methods for estimatiton of somatic-cell count in bovine milk. 2. N-acetyl-β-D-glucosaminidase test for routine estimation of somatic cell count in milk. Journal of Dairy Research 43: 491-494.

Kitchen, B.J., Kwee, W.S., Middleton, G. & Andrews, R.J., (1984b). Relationship between the level of N-acetyl-β-D-glucosaminidase (NAGase) in bovine milk and the presence of mastitis pathogens. J. Dairy Research 51:11-16.

Kitchen, B.J., Middleton, G. & Kwee, W.S. (1984a). N-acetyl-β-D-glucosaminidase (NAGase) levels in bulk herd milk. J. Dairy Research 51:227-232.

Kitchen, B.J., Middleton, G. and Salmon, M. (1978). Bovine milk N-acetyl-β-D-glucosaminidase and its significance in the detection of abnormal udder secretions. Journal of Dairy Research 45: 15-20.

Kitchen, B.J., Middleton, G., Durward, I.G., Andrews, R.J. & Salmon, M.C. (1980). Mastitis diagnostic tests to estimate mammary gland epithelial cell damage. J. Dairy Sci. 63:978.

Klebanoff, S.J. (1991). Myeloperoxidase:Occurrence and biology, In: Peroxidase in Chemistry and Biology J. Everse, K.E. Everse and A. Grisham, vol 1 CRC Press Inc. Boca Raton.

Klein, M.S., Almstetter, M.F., Schlamberger, G., Nurnberger, N., Dettmer, K., Oefner, P.J., Meyer, H.H.D., Wiedemann, S. and Gronwald, W. (2010). Nuclear magnetic resonance and mass spectrometry-based milk metabolomics in dairy cows during early and late lactation. Journal of Dairy Science 93: 1539-1550.

Klose, J. (1975). Protein mapping by combined isoelectric focusing and electrophoresis of mouse tissue. A novel approach to testing for induced point mutations in mammals. Humangenetik 26:231.

Knudsen, K. and Jensen, N.E. (1990). The diagnosis of subclinical mastitis by redefinition of the interquarter ratio and definition of the interquarter variation using analysis of variance methods. J. Dairy Res. 57:325.

Koj, A. (1974). Acute phase reactants. Their synthesis, turnover and biological significance. In Structure and Function of Plasma Proteins. Allison, A.C. (ed.), Vol. 1, p. 73. London: Plenum Press.

Köllisch, G.V., Lorenz, M.W., Kellner, R., Verhaert, P.D. and Hoffman, K.H. (2000). Structure elucidation and biological activity of an unusual adipokinetic hormone from corpora cardiac of the butterfly, Vanessa cardui. Eur. J. Biochem. 267: 5502-5508.

Korhonen, H., Marnila, P. and Gill, H.S. (2000). Milk immunoglobulins and complement factors. British J. Nutr., 84 Suppl 1: 75–80. 306

Kováč, G., Popelková, M., Tkáčiková, Ľ., Burdová, O. and Ihnát, O. (2007). Interrelationship between somatic cell count and acute phase proteins in serum and milk of dairy cows. Acta Vet Brno 76: 51-57.

Kremer, W.D.J., Noordhuizen-Stassen, E.N., Grommers, F.J., Daemen, A.J.J.M., Brand, A. and Burvenich, C. (1993). Blood polymorphonuclear leukocyte chemotaxis during experimental Escherichia coli bovine mastitis. Journal of Dairy Science 76: 2613-2618.

Kromker, V., Grabowski, N.T., Redetzky, R. and Hamann, J. (2001). Detection of mastitis using selected quarter milk parameters. 2nd Intl. Symp. Bovine Mastitis and Milk Quality. Vancouver, Canada, pp: 486-487.

Kuehl, F.A. and Egan, R.W. (1980)Prostaglandins, arachidonic acid and inflammation. Science 210: 978-984.

Kukovics, S., Molnar, A., Abraham, M. and Schuszter, T. (1996). Phenotypic correlation between somatic cell count and milk components. Allata.Takarm. 45: 205-215.

Labohm, R., Gotz, E., Luhofen, G., Hess, R.G. and Bostedt, H. (1998). Factors influencing the SCC in dairy cows. 1. Influence of bacteriological findings, stage and number of lactation. Milchwissenschaft 52: 63-66.

Lacasse, P., Farr, V.C., Davis, S.R. and Prosser, C.G. (1996). Local secretion of nitric oxide and the control of mammary blood flow. Journal of Dairy Science 79: 1369-1374.

Lacy-Hulbert, S. J., Woolford, M.W., Nicholas, G., Prosser, C.G. and Stelwagen, K. (1999). Effect of milking frequency and pasture intake on milk yield and composition of late lactation cows. J. Dairy Sci. 82:1232–1239.

Laevens, H., Deluyker, H., Schukken, Y.H., De Meulemeester, L., Vandermeersch, R., De Muelenaere, E. and De Kruif, A. (1997). Influence of parity and stage of lactation on the somatic cell count in bacteriologically negative dairy cows. Journal of Dairy Science 80: 3219-3226.

Lai, H., Tsao, J.H., Lu, Y.P., Lee, J.W., Zhao, X., Chien, F.L. and Mao, S.J.T. (2009). Neutrophils as one of the major haptoglobin sources in mastitis affected milk. Vet. Res. 40: 17.

Larsen, L.B., Hinz, K., Jorgensen, A.L.W., Moller, H.S., Wellnitz, O., Bruckmaier, R.M. and Kelly, A.L. (2010). Proteomic and peptidomic study of proteolysis in quarter milk after infusion with lipotechoic acid from Staphylococcus aureus. Journal of Dairy Science 93: 5613-5626. 307

Larsen, L.B., McSweeney, P.L.H., Hayes, M.G., Andersen, J.B., Ingvartsen, K.L. and Kelly, A.L. (2006). Variation in activity and heterogeneity of bovine milk proteases with stage of lactation and somatic cell count. International Dairy Journal 16: 1-8.

Larsen, L.B., Rasmussen, M.D., Bjerring, M. and Nielsen, J.H. (2004). Proteases and protein degradation in milk from cows infected with Streptococcus uberis. International Dairy Journal 14: 899-907.

Larsen, T., Rontved, C.M., Ingvartsen, K.L., Vels, L. and Bjerring, M. (2010). Enzyme activity and acute phase proteins in milk utilized as indicators of acute clinical E. coli LPS-induced mastitis. Animal 4: 1672-1679.

Larson, M.A., Weber, A., Weber, A.T. and McDonald, T.L. (2005). Differential expression and secretion of bovine serum amyloid A3 (SAA3) by mammary epithelial cells stimulated with prolactin or lipopolysaccharide. Veterinary Immunology and Immunopathology 107: 255-264.

Laville, R., Sayd, T., Terlouw, C., Chambon, C., Damon, M., Larzul, C., Le Roy, P., Glenisson, J. and Cherel, P. (2007). Comparison of sarcoplasmic proteomes between two groups of pig muscles selected for shear force of cooked meat. Journal of Agricultural and Food Chemistry 55: 5834-5841.

Le Bars, D. and Gripon, J.D. (1989). Specificity of plasmin towards bovine αs2-casein. Journal of Dairy Research 56: 817-821.

Le Gall, G., Colquhoun, I.J., Davis, A.L., Collins, G.J. and Verhoeyen, M.E. (2003). Metabolite profiling of tomato using 1H NMR spectroscopy as a tool to detect potential unintended effects following a genetic modification. Journal of Agricultural and Food Chemistry 51: 2447-2456.

Le Roux, Y., Laurent, F. and Moussaoui, F. (2003). Polymorphonuclear proteolytic activity and milk composition change. Vet. Res. 34: 629-645.

Le, A., Barton, L.D., Sanders, J.T. and Zhang, Q.A. (2011). Exploration of bovine milk proteome in colostral and mature whey using an ion-exchange approach. Journal of Proteome Research 10: 692- 704.

Lehtolainen, T., Rontved, C. and Pyorala, S. (2004). Serum amyloid A and TNF alpha in serum and milk during experimental endotoxin mastitis. Vet. Res. 35: 651-659.

Leitner, G., Shoshani, E., Krifucks, O., Chaffer, M. and Saran, A. (2000). Milk leucocyte populations in heifers free of udder infection. Journal of Veterinary Medicine Series B-Infectious Diseases and Veterinary Public Health 47:133-138. 308

Leonil, J. & Molle, D. (1991). A method for determination of macropeptide by cation-exchange fast protein liquid ehromatography and its use for following the action of chymosin in milk. Journal of Dairy Research 58: 321-328.

Levieux, D. and Venien, A. (1994). Rapid, sensitive two-site ELISA for detection of cows' milk in goats' or ewes' milk using monoclonal antibodies. Journal of Dairy Research 61: 91-99.

Li, J., Xu, X., Xu, C., Zhou, W., Zhang, K., Yu, H., Zhang, Y., Zheng, Y., Rees, H.H., Lai, R., Yang, D. and Wu, J. (2007). Anti-infection peptidomics of amphibian skin. Mol. Cell Proteomics 6: 882– 894.

Liang, C. (1957). The formation of complexes between haemoglobin and plasma proteins in a variety of animals. Biochem. J. 66: 552-558.

Liang, J. and Sipe, J.D. (1995). Recombinant human serum amyloid A (apoSAAp) binds cholesterol and modulates cholesterol flux. Journal of Lipid Research 36: 37-46.

Liao, Y.L., Alvarado, R., Phinney, B. and Lonnerdal, B. (2011). Proteomic characterization of human milk whey proteins during a twelve-month lactation period. Journal of Proteome Research 10: 1746- 1754.

Lilley, K.S., Razzaq, A. and Dupree, P. (2002). Two-dimensional gel electrophoresis: recent advances in sample preparation, detection and quantitation. Current Opinion in Chemical Biology 6: 46-50.

Lindmark-Månsson, H. and Åkesson, B. (2000). Antioxidative factors in milk. British Journal of Nutrition 84: 103–110.

Lindmark-Mansson, H., Branning, C., Alden, G. and Paulsson, M. (2006). Relationship between somatic cell count, individual leukocyte populations and milk components in bovine udder quarter milk. International Dairy Journal 16: 717-727.

Linke, R.P., Bock, V., Valet, G. and Rothe, G. (1991). Inhibition of the oxidative burst response of N formyl peptide-stimulated neutrophils by serum amyloid A protein. Biochem Biophys Res Commun 176: 1100–1105.

Linzell, J.L. and Peaker, M. (1971). Early detection of mastitis. Veterinary Record 14: 393-394.

Linzell, J.L. and Peaker, M. (1972). Day-to-day variations in milk composition in the goat and cow as a guide to the detection of subclinical mastitis. Br. Vet. J. 128:284.

Linzell, J.L., Peaker, M. and Rowell, J.G. (1974). Electrical conductivity of foremilk for detecting subclinical mastitis in cows. J. Agric. Sci. (Camb) 83:309. 309

Lippolis, J. D. and Reinhardt T. A. (2005). Proteomic survey of bovine neutrophils. Vet. Immunol. Immunopathol. 103: 53–65.

Lippolis, J. D., White, F. M., Marto, J. A., Luckey, C. J., Bullock, T. N., Shabanowitz, J., Hunt, D. F. and Engelhard, V. H.. (2002). Analysis of MHC Class II antigen processing by quantitation of peptides that constitute nested sets. J. Immunol. 169:5089–5097.

Lippolis, J.D. & Reinhardt, T.A. (2010). Utility, limitations and promise of proteomics in animal science. Veterinary Immunology Immunopathology 138: 241-251.

Lippolis, J.D. and Reinhardt, T.A. (2008). Centennial Paper: Proteomics in Animal Science. Journal of Animal Science 86: 2430-2441.

Lippolis, J.D., Peterson-Burch, B.D. and Reinhardt, T.A. (2006a). Differential expression analysis of proteins from neutrophils in the periparturient period and neutrophils from dexamethasone-treated dairy cows. Veterinary Immunology and Immunopathology 111: 149-164.

Lippolis, J.D., Reinhardt, T.A., Goff, J.P. and Horst, R.L. (2006b). Neutrophil extracellular trap formation by bovine neutrophils is not inhibited by milk. Veterinary Immunology and Immunopathology 113: 248-255.

Little, T.W.A., Herbert, C.N. and Forbes, D. (1968). Electrical conductivity and the leucocyte count of bovine milk. Veterinary Record 82:431-433.

Lock, A.L. and Garnsworthy, P.C. (2003). Seasonal variation in milk conjugated linoleic acid and Δ9- desaturase activity in dairy cows. Livestock Production Science 79: 47-59.

Ly, L., Valerie, C. and Wasinger, V.C. (2008). Peptide enrichment and protein fractionation using selective electrophoresis. Proteomics 8: 4197-4208.

Ma, Y., Ryan, C., Barbano, D.M., Galton, D.M., Rudan, M.A. and Boor, K.J. (2000). Effects of somatic cell count on quality and shelf-life of pasteurized fluid milk. J. Dairy Sci., 83: 264-274.

Maatje, K., Rossing, W., Garssen, G.J. and Pluygers, H.G. (1983). Automation of electrical conductivity measurements during milking. Page 89 in Proc. 3rd Symp. Automation in Dairying, Wageningen, Neth.

Maddox, J.F., Reddy, C.C., Eberhart, R.J. and Scholz, R.W. (1991). Dietary selenium effects on milk eicosanoid concentration in dairy cows during coliform mastitis. Prostaglandins 42: 369-378.

Madureira, A.R., Pereira, C.I., Gomes, A.M.P., Pintado, M.E. and Malcata, F.X. (2007). Bovine whey proteins-Overview on their main biological properties. Food Research International 40: 1197-1211. 310

Maiti, S.K., Sharma, N. and Koley, K.M. (2004). Efficacy of post milking herbal tea dip in the control of mastitis in cows. Vet. Practitioner 5:105-106.

Makimura, S. and Suzuki, N. (1982). Quantitative determination of bovine serum haptoglobin and its elevation in some inflammatory disease. Jpn J Vet Sci 44: 15–21.

Mamone, G., Caira, S., Garro, G., Nicolai, A., Ferranti, P., Picariello, G., Malorni, A., Chianese, L. and Addeo, F. (2003). Casein phosphoproteome: Identification of phsophoproteins by combined mass spectrometry and two-dimensional gel electrophoresis. Electrophoresis 24: 2824-2837.

Man, E.H. and Bada, J.L. (1987). Dietary D-amino acids. Ann. Rev. Nutr. 7: 209-225.

Manso, M.A., Leonil, J., Jan, G. and Gagnaire, V. (2005). Application of proteomics to the characterisation of milk and dairy products. International Dairy Journal 15: 845-855.

Marhaug, G., Permin, H. and Husby, G. (1983). Amyloid-related serum protein (SAA) as an indicator of lung infection in cystic fibrosis. Acta Paediatr. Scand. 72: 861.

Marino, R., Considine, T., Sevi, A., McSweeney, P.L.H. and Kelly, A.L. (2005). Contribution of proteolytic activity associated with somatic cells in milk to cheese ripening. International Dairy Journal 15: 1026-1033.

Marschke, R.J. and Kitchen, B.J. (1984). Glucose levels in normal and mastitic milk. Journal of Dairy Research 51: 233-237.

Marshall, R.T. and Edmondson, J.E. (1962). Value of California mastitis test records to the practitioner. J. Am. Vet. Med. Ass. 140: 45-49.

Martens, H., Kohler, A., Afseth, N.K., Wold, J.P., Hersleth, M., Berget, I., Adnoy, T., Skaugen, M., Isaksson, T., Vegarud, G., Criscione, A., Mevik, B.H., Frost, M.B., Randby, A.T., Prestlokken, E., Berg, P., Kent, M., Lien, S. and Omholt, S.W. (2007). High-throughput measurements for functional genomics of milk. Journal of Animal and Feed Sciences 16: 172-189.

Maruyama, S. and Suzuki, H. (1982). A peptide inhibitor of angiotensin-1 converting enzyme in the tryptic hydrolysate of casein. Agricultural and Biological Chemistry 46: 1393-1394.

Massart-Leen, A-M., Burvenich, C. and Massart, D.L. (1994). Triacylglycerol fatty acid composition of milk from periparturient cows during acute Escherichia coli mastitis. Journal of Dairy Research 61: 191-199.

Mather, I. H. (2000). A review and proposed nomenclature for major proteins of the milk-fat globule membrane. J. Dairy Sci. 83: 203–247. 311

Mattila, T., Pyörälä, S. and Sandholm, M. (1986). Comparison of milk antitrypsin, albumin, N-acetyl- β-D-glucosaminidase, somatic cells and bacteriological analysis as indicators of bovine subclinical mastitis. Vet. Res. Comm. 10: 113-124.

Mazur, D. and Eugeniusz-Herbut, J.W. (2006). Infrared thermography as a diagnostic method. Roczniki Naukowe Zootechniki 33: 171–181.

McClenahan, D. J., Sotos, J.P. and Czuprynski, C.J. (2005). Cytokine response of bovine mammary gland epithelial cells to Escherichia coli, coliform culture filtrate, or lipopolysaccharide. Am. J. Vet. Res. 66:1590–1597.

McDonald, T.L., Larson, M.A., Mack, D.R. and Weber, A. (2001). Elevated extrahepatic expression and secretion of mammary-associated serum amyloid A3 (M-SAA3) into colostrums. Veterinary Immunology and Immunopathology 83: 203-211.

McDonald, T.L., Weber, A. and Smith, J.W. (1991). A monoclonal antibody sandwich immunoassay for serum amyloid A (SAA) protein. Journal of Immunological Methods 144: 149-155.

McFadden, T.B., Akers, R.M. and Beal, W.E. (1988). Milk protein secretion by explants of prepubertal bull mammary tissue: Breed difference. J. Dairy Sci. 71: 2904-2914.

McGavin, D. and Zachary, J.F. (2007). Pathologic Basis of Veterinary Disease. 4th ed. Mosby Elsevier, St. Louis, MO.

Mcmeekin, T.L. (1970). Milk proteins in retrospect. Milk proteins, chemistry and molecular biology Academic Press: New York, N.Y., U.S.A., London, England 3-15.

McNair, J., Elliott, C.T. and Mackie, D.P. (1995). Development of a sensitive and specific time resolved fluorimetric immunoassay for the bovine acute phase protein haptoglobin (Hp). Journal of Immunological Methods 184: 199-205.

McNair, J., Kennedy, G., Bryson, D.G., Reilly, G.A.C., McDowell, S.W.J. and Mackie, D.P. (1997). Evaluation of a competitive immunoassay for the detection of bovine haptoglobin. Research in Veterinary Science 63: 145-149.

Meek, R.L. and Benditt, E.P. (1986). Amyloid A gene family expression in different mouse tissues. J. Exp. Med. 164: 2006–2017.

Meisel, H. and FitzGerald, R.J. (2003). Biofunctional peptides from milk proteins: Mineral binding and cytomodulatory effects. Current Pharmaceutical Design 9: 1289-1295. 312

Meisel, H. and Schlimme, E. (1994). Inhibitors of angiotensin converting enzyme derived from bovine casein (casokinins). In: b-casomorphins and related peptides Recent Developments, eds V. Brantl and H. Teschemacher. Weinheim, New York, VCH pp. 27-33.

Melendez, P., Marin, M.P., Robles, J., Rios, C., Duchens, M. and Archbald, L. (2009). Relationship between serum nonesterified fatty acids at calving and the incidence of periparturient diseases in Holstein dairy cows. Theriogenology 72: 826-833.

Mellenberger, R. (2001).California Mastitis Test (CMT). An invaluable tool for managing mastitis. Dept. Of Animal Sciences, Michigan State University 1-9.

Mellors, A. (1968). Bovine N-acetylglucosaminidase in bovine milk. Can. J. Biochem. 46: 451-455.

Michaelidou, A. and Steijns, J. (2006). Nutritional and technological aspects of minor bioactive components in milk and whey: Growth factors, vitamins and nucleotides. International Dairy Journal 16: 1421-1426.

Middleton, J.R. and Fox, L.K. (2001). Therapeutic cessation of lactation of Staphylococcus aureus- infected mammary quarters. J Dairy Sci. 84:1976–1978.

Middleton, J.R., Hardin, D., Steevens, B., Randle, R. and Tyler, J.W. (2004). Use of somatic cell counts and California Mastitis Test results from individual quarter milk samples to detect subclinical intramammary infection in dairy cattle from a herd with a high bulk tank somatic cell count. JAVMA 224:419-423.

Milica, K.F., Jelka, S., Marija, S.P., Jasmina, D.M., Milijana, K., Zora, M. and Tatjana, B. (2010). Acute phase protein response in cows with Staphylococcus aureus subclinical mastitis. Acta Veterinaria (Beograd) 60: 205-216.

Miller, D.D. and Kearns, J.V. (1966). Effectiveness of the California mastitis test as a measurement of the leucocyte content of quarter samples of milk. Journal of Dairy Science 50: 683-686.

Miller, G.Y., Bartlett, P.C., Lance, S.E., Anderson, J. and Heider, L.E. (1993). Costs of clinical mastitis and mastitis prevention in dairy herds. Journal of the American Veterinary Medical Association 202: 1230-1236.

Miller, R. H. and Paape, M.J. (1988). Effects of parity, bacteriological status, stage of lactation, and dry period on N-acetyl-β-D-glucosaminidase activity of milk and dry secretions. J. Dairy Sci., 71: 2508-2512. 313

Milner, P., Page, K.L., Walton, A.W. and Hillerton, J.E. (1996). Detection of clinical mastitis by changes in electrical conductivity of foremilk before visible changes in milk. J. Dairy Sci 79:83-86.

Mitchell, G.B., Clark, M.E. and Caswell, J.L. (2007). Alterations in the bovine bronchoalveolar lavage proteome induced by dexamethasone. Veterinary Immunoly and Immunopathology 118: 283- 293.

Mitton, B. and Kranias, E.G., Proteomics: (2003).Its potential in the post-genome era. Hellenic Journal of Cardiology 44: 301-307.

Möhring, T., Kellman, M., Jürgens, M. and Schrader, M. (2005). Top-down identification of endogenous peptides up to 9 kDa in cerebrospinal fluid and brain tissue by nanoelectrospray quadrupole time-of-flight tandem mass spectrometry. Journal of Mass Spectrometry 40: 214-226.

Molenaar, A., Rajan, G., Pearson, M., Miles, M., Petrova, R., Davis, S. and Stelwagen, K. (2002). A serum amyloid protein homologue is expressed by the mammary gland in a similar pattern to Lactoferrin, Proc. Third Eur. Colloq. Food Safety Acute Phase Prot. Doorn, Utrecht, The Netherlands Page 42.

Molenaar, A.J., Harris, D.P., Rajan, G.H., Pearson, M.L., Callaghan, M.R., Sommer, L., Farr, V.C., Oden, K.E., Miles, M.C., Petrova, R.S., Good, L.L., Singh, K., McLaren, R.D., Prosser, C.G., Kim, K.S., Wieliczko, R.J., Dines, M.H., Johannessen, K.M., Grigor, M.R., Davis, S. and Stelwagen K. (2009). The acute-phase protein serum amyloid A3 is expressed in the bovine mammary gland and plays a role in host defence. Biomarkers 14: 26-37.

Molina, E., Frutos, M. and Ramos, M. (2000). Capillary electrophoresis characterization of the casein fraction of cheeses made from cows’, ewes’ and goats’ milks. Journal of Dairy Research 67: 209-16.

Mollenkopfa, H. J., Grode, L., Mattow, J., Stein, M., Mann, P., Knapp, B., Ulmer, J. and Kaufmann, S.H.E. (2004). Application of Mycobacterial proteomics to vaccine design: Improved protection by Micobacterium bovis BCG Prime-Rv3407 DNA boost vaccination against tuberculosis. Infection and Immunity 72: 6471-6479.

Moncada, S., Palmer, R.M. and Higgs, E.A. (1991). Nitric oxide: physiology, pathophysiology, and pharmacology. Pharmacol Rev. 43: 109–142.

Moore, R.E., Kirwan, J., Doherty, M.K. and Whitfield, P.D. (2007). Biomarker discovery in animal health and disease: the application of post-genomic technologies. Biomarker Insights 2: 185-196. 314

Morimatsu, M., Sarikaputi, M., Syuto, B., Saito, M., Yamamoto, S. and Naiki, M. (1992). Bovine haptoglobin: Single radial immunodiffusion assay of its polymeric forms and dramatic rise in acute- phase sera. Veterinary Immunology and Immunopathology 33: 365-372.

Morimatsu, M., Syuto, B., Shimada, N., Fujinaga, T., Yamamoto, S., Saito, M. and Naiki, M. (1991). Isolation and characterization of bovine haptoglobin from acute phase sera. J. Biol. Chem. 266: 11833–11837.

Moyer, S.C., Budnik, B.A., Pittman, J.L., Costello, C.E. and O’Connor, P. B., (2003). Attomole peptide analysis by high-pressure matrix-assisted laser desorption/ionization Fourier transform mass spectrometry. Anal. Chem. 75:6449–6454.

Moyes, K.M., Larsen, T., Friggens, N.C., Drackley, J.K. and Ingvartsen, K.L. (2009). Identification of potential markers in blood for the development of subclinical and clinical mastitis in dairy cattle at parturition and during early lactation. Journal of Dairy Science 92: 5419-5428.

Mullally, M. M., Meisel, H. and FitzGerald, R. J. (1996). Synthetic peptides corresponding to a- lactalbumin and b-lactoglobulin sequences with angiotensin-I-converting enzyme inhibitory activity. Biological Chemistry Hoppe-Seyler 377: 259-260.

Muller, L., Bartak, P., Bednar, P., Frysova, I., Sevcik, J. and Lemr, K. (2008). Capillary electrophoresis-mass spectrometry- a fast and reliable tool for the monitoring of milk adulteration. Electrophoresis 29: 2088-2093.

Murakami, K., Lagarde, M. and Yuki, Y. (1998). Identification of minor proteins of human colostrums and mature milk by two-dimensional electrophoresis. Electrophoresis 19: 2521-2527.

Murata, H., and Miyamoto, T. (1993). Bovine haptoglobin as a possible immunomodulator in the sera of transported calves. Br. Vet. J. 149: 277–283.

Murata, H., Shimada, N. and Yoshioka, M. (2004). Current research on acute phase proteins in veterinary diagnosis: an overview. The Veterinary Journal 168: 28-40.

Murgiano, L., Timperio, A.M., Zolla, L., Bongiorni, S., Valentini, A. and Pariset, L. (2009). Comparison of milk fat globule membrane (MFGM) proteins of Chianina and Holstein cattle breed milk samples through proteomics methods. Nutrients 1: 302-315.

Murphy, J. M. (1942). Further Observations on the Modified Whitcside Test for the Detection of Chronic Bovine Mastitis. Cornell Vet. 32: 439. 315

Murphy, J.M. & Stuart, O.M., (1954). The syringe collection of milk via teat-wall punctures as an aid in the study of artificial infection of the bovine mammary gland. Cornell Veterinarian 44:501-511.

Murphy, J.M. and Hanson, J.J. (1941). A Modified Whiteside Test for the Detection of Chronic Bovine Mastitis. Cornell Vet. 31:47.

Murphy, S.C., Cranker, K., Senyk, G.F., Barbano, D.M., Saeman, A.I. and Galton, D.M. (1989). Influence of bovine mastitis on lipolysis and proteolysis in milk. Journal of Dairy Science 72: 620- 626.

Nafikov, R.A. and Beitz, D.C. (2007). Carbohydrate and lipid metabolism in farm animals. J. Nutr. 137: 702-705.

Nagahata, H., Saito, S. and Noda, H. (1987). Changes in N-Acetyl-β-D-Glucosaminidase and β- Glucuronidase Activities in Milk during Bovine Mastitis. Can J Vet Res 51:126-134.

Nakagawa-Tosa, N., Morimatsu, M., Kawasaki, M., Nakatsuji, H., Syuto, B. and Saito, M. (1995) Stimulation of haptoglobin synthesis by interleukin-6 and tumor necrosis factor, but not by interleukin-1, in bovine primary cultured hepatocytes. Journal of Veterinary Medical Science 57: 219–223.

Napel, T., de Haas, Y. and de Jong, et al. (2009). Characterization of distributions of somatic cell counts. Journal of Dairy Science 92:1253-1264.

Nauseef, W.M., Olsen, I. and Amljots, K. (1988). Myeloperoxidase. Eur. J. Haematol. 40:97-110.

Nazifi, S., Khoshvaghti, A. and Gheisari, H.R. (2008). Evaluation of serum and milk amyloid A in some inflammatory diseases of cattle. Iranian Journal of Veterinary Research 9: 222-226.

Needs, E. C. and Anderson, M. (1984). Lipid composition of milks from cows with experimentally induced mastitis. J. Dairy Res. 51: 239.

Neuhaus, O.W. & Sogoian, V. P. (1961). The presence of haptoglobin in synovial fluid. Nature 192: 558-559.

Neuhoff, V., Arold, N., Taube, D. and Ehrhardt, W. (1988). Improved staining of proteins in polyacrylamide gels including isoelectric focusing gels with clear background at nanogram sensitivity using Coomassie Brilliant Blue G-250 and R-250. Electrophoresis 9: 255–262.

Neverova, I. and Van Eyk, J.E. (2002). Application of reversed phase high performance liquid chromatography for subproteomic analysis of cardiac muscle. Proteomics. Special Issue: Proceedings 316 of the Britain Electrophoresis Society Meeting York, UK, April 2001- From Biology to Pathology-the Proteomics Perspective. 2: 22-31.

Nielen, M., Deluyker, H., Schukken, Y.H. & Brand, A. (1992). Electrical Conductivity of Milk: Measurement, Modifiers and Meta Analysis of Mastitis Detection Performance. J. Dairy Sci 75:606- 614.

Nielen, M., Schukken, Y.H., Brand, A., Haring, S. and Ferwerda-van Zonneveld, R.T. (1995). Comparison of some analysis techniques for on-line clinical mastitis detection. Journal of Dairy Science 78: 1050-1061.

Nielsen, B.H., Jacobsen, S., Andersen, P.H., Niewold, T.A. & Heegaard, P.M.H. (2004). Acute phase protein concentrations in serum and milk from healthy cows, cows with clinical mastitis and cows with extramammary inflammatory conditions. Veterinary Record 154: 361–365.

Nielsen, J. and Jewett, M.C. (2008). Minireview: Impact of systems biology on metabolic engineering of Saccharomyces cerevisiae. FEMS Yeast Res 8: 122-131.

Nielsen, L., Toft, N., Eckersall, P.D., Mellor, D.J. and Morris, J.S. (2007). Serum C-reactive protein concentration as an indicator of remission status in dogs with multicentric lymphoma, Journal of Veterinary Internal Medicine 21: 1231–1236.

Nikkhah, A., Plaizier, J.C., Einarson, M.S., Berry, R.J., Scott, S.L. and Kennedy, A.D. (2005). Infrared thermography and visual examination of hooves of dairy cows in two stages of lactation. J. Dairy Sci. 88:2749–2753.

Norberg, E., Hogeveen, H., Korsgaard, I.R., Friggens, N.C., Sloth, K.H.M.N. and Lovendahl, P. (2003). Electrical conductivity of milk: Ability to predict mastitis status. J. Dairy Sci 87:1009-1107.

Nyman, A.K., Emanuelson, U., Holtenius, K., Ingvartsen, K.L., Larsen, T. and Persson, W.K. (2008). Metabolites and immune variables associated with somatic cell counts of primiparous dairy cows. Journal of Dairy Science 91: 2996-3009.

O’Donnell, R., Holland, J.W., Deeth, H.C. & Alewood, P. (2004). Review: Milk proteomics. International Dairy Journal 14: 1013-1023.

O’Farrell, P.H. (1975). High resolution two-dimensional electrophoresis of proteins. J. Biol.Chem. 250: 4007-4021. 317

O’Mahony, M.C., Healy, A.M., Harte, D., Walshe, K.G., Torgerson, P.R. and Doherty, M.L. (2006). Milk amyloid A: Correlation with cellular indices of mammary inflammation in cows with normal and raised serum amyloid A. Research in Veterinary Science 80: 155-161.

Obara, Y. & Komatsu, M. (1984). Relationship Between N-Acetyl-β-D-Glucosaminidase Activity and Cell Count, Lactose, Chloride, or Lactoferrin in Cow Milk. J. Dairy Sci 67:1043-1046.

Ohtsuka, H., Kudo, K., Mori, K., Nagai, F., Hatsugaya, A., Tajima, M., Tamura, K., Hoshi, F., Koiwa, M. and Kawamura, S. (2001). Acute phase response in naturally occurring coliform mastitis. Journal of Veterinary Medical Science 63: 675–678.

Oliver, S.G., Winson, M.K., Kell, D.B. and Baganz, R. (1998). Systematic functional analysis of the yeast genome. Trends Biotechnol. 16: 373-378.

Oliver, S.P. and Sordillo, L.M. (1988). Udder health in the periparturient period. Journal of Dairy Science 71: 2584-2606.

Ong, S.E. and Mann, M. (2005). Mass spectrometry-based proteomics turns quantitative. Nat Chem Biol 1:252–262.

Ong, S.E., Blagoev, B., Kratchmarova, I., Kristensen, D.B., Steen, H., Pandey, A. and Mann, M. (2002). Stable isotope labelling by amino acids in cell cultre, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics 1: 376-386.

Onyango, C.M., Marchant, J.A., Lake, J.R. and Stanbridge, D.A. (1988). A low maintenance conductivity sensor for detecting mastitis. J. Agric. Eng. Res. 40:215.

Oosterkamp, A.J., van der Hoeven, R., Glassgen, W., Konig, B., Tjaden, U.R., van der Greef, J. & Irth, H. (1998). Gradient reversed-phase liquid chromatography coupled on-line to receptor-affinity detection based on the urokinase receptor. J. Chromatogr. B. 715: 331–338.

Oshima, M. (1977). Detection of abnormal quarter milk by quarter difference of electrical conductivity and its theoretical basis. Jarq-Japan Agricultural Research Quarterly 11: 239-245.

Owen, C.A. and Campbell, E.J. (1999). The cell biology of leukocyte-mediated proteolysis. Journal of Leukocyte Biology 65: 137-150.

Owen, J.A., Better, F.C. and Hoban, J. (1960). A simple method for the determination of serum haptoglobins. J. Clin. Pathol. 13: 163-164.

Paape, M.J., Bannerman, D.D., Zhao, X. and Lee, J.W. (2003). The bovine neutrophil: Structure and function in blood and milk. Vet. Res. 34: 597-627. 318

Palmer, D.J., Kelly, V.C., Smit, A-M., Knight, C.G. and Cooper, G.J. (2006). Human colostrums: Identification of minor proteins in the aqueous phase by proteomics. Proteomics 6: 2208-2216.

Panndorf, H., Richter, H. and Dittrich, B. (1976). Haptoglobin in domestic mammals V. Plasma haptoglobin level in cattle under pathological conditions. Arch. Exp. Vet. 30: 193-202.

Pareek, R., Wellnitz, O., Dorp, R.V., Burton, J. and Kerr, D. (2005). Immunorelevant gene expression in LPS-challenged bovine mammary epithelial cells. J Appl Genet 46: 171-177.

Parker, F., Migliore-Samour, D. and Floc'h, F. (1984). Immunostimulating hexapeptide from human casein, amino acid sequence, synthesis and biological properties. European Journal of Biochemistry 145: 677-682.

Peaker, M. (1978). The electrical conductivity of milk for the detection of subclinical mastitis in cows: comparison of various methods of handling conductivity data with the use of cell counts and bacteriological examination. Br. Vet. J. 134:308.

Pedersen, L.H., Aalbæk, B., Røntved, C.M., Ingvartsen, K.L., Sorensen, N.S., Heegard, P.M.H. and Jensen, H.E. (2003). Early pathogenesis and inflammatory response in experimental bovine mastitis due to Streptococcus uberis. J. Comp. Path. 128: 156-164.

Pepys, M.B. and Baltz, M.L. (1983). Acute phase proteins with special reference to C-reactive protein and related proteins (pentaxins) and serum amyloid A protein. Adv. Immunol. 34: 141–212.

Perdigon, G., Medici, M., Cecilia, M., Nader, deMacias, M.E., Haedo, R., Oliver, G. and Pesce deRuizholgado, A.A. (1986). Significance of the Presence of Bovine Milk β-Glucuronidase in Mastitis Detection J Dairy Sci 69:27-31.

Peter, A. T., Clark, P.W., van Roekel, D.E., Luker, C.W., Gaines, J.D. and Bosu, W.T. (1990). Temporal changes in metabolites of prostanoids in milk of heifers after intramammary infusion of Escherichia coli organisms. Prostaglandins 39:451–457.

Peters, Jr. T. (1977). Serum albumin: recent progress in the understanding of its structure and biosynthesis. Clinical Chemistry 23: 5-12.

Petricoin, E.F., Belluco, C., Araujo, R.P. and Liotta, L.A. (2006). The blood peptidome: a higher dimension of information content for cancer biomarker discovery. Nature Reviews Cancer 6: 961- 967.

Pingali, A., McGuninnes, B., Keshishian, H., Fei-Wu, J., Varady, L. and Regnier, F. (1996). Peptides as affinity surfaces for protein purification. Journal of Molecular Recognition 9: 426-432. 319

Pinotti, L., Baldi, A. and Dell’Orto, V. (2002). Comparative mammalian choline metabolism with emphasis on the high-yielding dairy cow. Nutrition Research Reviews 15: 315-331.

Plath, A., Einspanier, R., Gabler, C., Peters, F., Sinowatz, F., Gospodarowicz, D. and Schams, D. (1998). Expression and localization of members of the fibroblast growth factor family in bovine mammary gland. Journal of Dairy Science 81: 2604-2613.

Pohn, G., Albert, C., Csapo-Kiss, Z. and Csapo, J. (2009). Influence of mastitis on D-amino acid content of milk. Acta Univ. Sapientiaem, Alimentaria 2: 31-43.

Polat, B., Colak, A., Cengiz, M., Yanmaz, L.E., Oral, H., Bastan, A., Kaya, S. & Hayirli, A. (2010). Sensitivity and specificity of infrared thermography in detection of subclinical mastitis in dairy cows. J. Dairy Sci. 93:3525-3532.

Politis, L., Lachance, E., Block, E. and Turner, J.D. (1989). Plasmin and plasminogen in bovine milk: A relationship with involution? Journal of Dairy Science 72: 900-906.

Polson, C., Sarkar, P., Incledon, B., Raguvan, V. and Grant, R. (2003). Optimization of protein precipitation based upon effectiveness of protein removal and ionization effect in liquid chromatography-tandem mass spectrometry. Journal of Chromatography B 785: 263-275.

Puckett, H.B., Fernando, R.S., Spahr, S.L. and Rodda, E.D. (1984). A system for analysis of milk production and conductivity data in real time. Agricultural electronics-1983 and beyond. Volume II: Controlled environments, livestock production systems, materials handling and processing. Page 508 in Proc. Natl. Conf. December 11-13, 1983, Chicago Proc. Am. Soc. Agric. Eng., St. Joseph, MI.

Pyorala, S. (2003). Indicators of inflammation in the diagnosis of mastitis. Vet.Res 34:565-578.

Pyörälä, S. and Pyörälä, E. (1997). Accuracy of methods using somatic cell counts and N-acetyl-β-D- glucosaminidase activity in milk to assess the bacteriological cure of bovine clinical mastitis. J. Dairy Sci. 80: 2820-2825.

Pyörälä, S., Hovinen, M., Simojoki, H., Fitzpatrick, J., Eckersall, P.D. and Orro, T. (2011). Acute phase proteins in milk in naturally acquired bovine mastitis caused by different pathogens. Veterinary Record 168: 535.

Qiao, F., Trout, D.R., Xiao, C. and Cant, J.P. (2005). Kinetics of glucose transport and sequestration in lactating bovine mammary gland measured in vivo with a paired indicator/nutrient dilution technique. Journal of Applied Physiology 99: 799-806. 320

Quaranta, S., Giuffrida, M.G., Cavaletto, M., Giunta, C., Godovac-Zimmermann, J., Cañas, B., Fabris, C., Bertino, E., Mombrò, M. and Conti, A. (2001). Human proteome enhancement: high- recovery method and improved two-dimensional map of colostral globule membrane proteins. Electrophoresis 22: 1810-1818.

Rabilloud, T. (2002). Two-dimensional gel electrophoresis in proteomics: Old, old fashioned, but it still climbs up the mountains. Proteomics 2: 3–10.

Radostits, O.M., Blood, D.C. & Gay, C.C. (Eds) (1994). Reference laboratory values. Veterinary Medicine, p. 1726. London: Bailliere Tindall.

Radostits, O.M., Blood, D.C., Gay, C.C., Blood, D.C. and Hinchkliff, K.W. (2000). Veterinary Medicine. 9th Edn, ELBS-Bailliere Tindal, London pp: 563-618.

Radsovevich, T.J., Reinhardt, T.A., Lippolis, J.D., Bannantine, J.P. and Stabel, J.R. (2007). Proteome and differential expression analysis of membrane and cytosolic proteins from Mycobacterium avium subsp. Paratuberculosis strains K-10 and 187. Journal of Bacteriology 189: 1109-1117.

Ramadori, G., Sipe, J.D. and Colten, H.R. (1985). Expression and regulation of the murine serum amyloid A (SAA) gene in extrahepatic sites. J Immunol. 135: 3645–3647.

Ramström, M. and Bergquist, J. (2004). Miniaturized proteomics and peptidomics using capillary liquid separation and high resolution mass spectrometry. FEBS Letters 567: 92-95.

Randolph, H.E. and Erwin, R.E. (1974). Influence of mastitis on properties of milk×Fatty acid composition. J. Dairy Sci. 57: 865–868.

Raulo, S.M., Sorsa, T.,Tervahartiala, T., Latvanen, T., Pirilä, E., Hirvonen, J. and Maisi, P. (2002). Increase in milk metalloproteinase activity and vascular permeability in bovine endotoxin-induced and naturally occurring Escherichia coli mastitis. Vet. Immunol. Immunopathol. 85: 137–145.

Rechinger, K.B., Siegumfeldt, H., Svendsen, I., and Jakobsen, M. (2000). “Early” protein synthesis of Lactobacillus delbrueckii ssp. bulgaricus in milk revealed by [35S] methionine labeling and two- dimensional gel electrophoresis. Electrophoresis 21: 2660–2669.

Recio, I., Amigo, L., Ramos, M., and Lopez-Fandino, R. (1997). Application of capillary electrophoresis to the study of proteolysis of caseins. Journal of Dairy Research 64: 221-230.

Reddy, L.V., Choudhuri, P.C., & Hamza, P.A. (1998). Sensitivity, specificity and predictive values of various indirect tests in the diagnosis of subclinical mastitis. Indian Veterinary Journal 75:1004– 1005. 321

Reichmuth, J. (1975). Somatic cell counting - interpretation of results. In: Proc. of Sem. On Mast. Cont. 1975, IDF Doc. 85. 93-109.

Reinhardt, T.A. and Lippolis, J.D. (2006). Bovine milk fat globule membrane proteome. Journal of Dairy Research 73: 406-416.

Reinhardt, T.A. and Lippolis, J.D. (2008). Developmental changes in the Milk Fat Globule Membrane Proteome during the Transition from Colostrum to Milk. Journal of Dairy Science 91: 2307-2318.

Rice, D.N. (1981). G81-556. Using the California Mastitis Test (CMT) to detect subclinical mastitis. University of Nebraska-Lincoln.

Richter, H. (1974). Haptoglobin bei Haussaugetieren. III. Der Haptoglobingehalt im Blutplasma und Serum von Wiederkauem und Schweinen unter vershiedenen physiologischen Bedingungen. Arch. Exp. Veterinaermed. 28: 505-519.

Rifai, N., Gillette, M.A. and Carr, S.A. (2006). Protein biomarker discovery and validation: the long and uncertain path to clinical utility. Nature Biotechnology 24: 971-983.

Roepstorff, P. & Fohlman, J. (1984). Proposal for a commo nomenclature for sequence ions in mass spectra of peptides. Biomedical Mass Spectrometry 11: 601.

Rook, J.A.F., Storry, J.E. and Wheelock, J.V. (1965). Plasma glucose and acetate and milk secretion in ruminant. Journal of Dairy Science 48: 745-747.

Rossbacher, J., Wagner, L. and Pasternack, M.S. (1999). Inhibitory effect of haptoglobin on granulocyte chemotaxis, phagocytosis and bactericidal activity. Scand. J. Immunol. 50:399–404.

Rowland, S.J. (1938). The determination of the nitrogen distribution in milk. Journal of Dairy Research 9: 42-46.

Ruegg, P.L. and Reinemann, D.J. (2002). Milk quality and mastitis tests. 35th Annual Conference of the American-Association-of-Bovine-Practitioners. Bovine Practitioner 36:41-54.

Saad, A.M. and Ostensson, K. (1990). Flow cytofluorometric studies on the alteration of leukocyte populations in blood and milk during endotoxin-induced mastitis in cows. Am. J. Vet. Res. 51:1603– 1607.

Safi, S., Khoshvaghti, A., Jafarzadeh, S.R., Bolourchi, M. and Nowrouzian, I. (2009). Acute phase proteins in the diagnosis of bovine subclinical mastitis. Veterinary Clinical Pathology 38: 471-476. 322

Saghatelian, A., Jessani, N., Joseph, A., Humphrey, M. and Cravatt, B.F. (2004). Activity-based probes for the proteomic profiling of metalloproteases. Proceedings of the National Academy of Sciences of the USA 101: 10000-10005.

Salih, A.M.A. and Anderson, M. (1979). Observations on the influence of high cell count on lipolysis in bovine milk. Journal of Dairy Research 46: 453–462.

Salonen, M., Hirvonen, J., Pyorala, S., Sankari, S. and Sandholm, M. (1996). Quantitative determination of bovine serum haptoglobin in experimentally induced Escherichia coli mastitis. Research in Veterinary Science 60: 88-91.

Sandholm, M., Honkanen-Buzalski, T. and Kangasniemi, R. (1984). Milk trypsin-inhibitor capacity as an indicator of bovine mastitis-a novel principle which can be automated. Journal of Dairy Research 51: 1-9.

Sargeant, J.M., Leslie, K.E., Shirley, J.E., Pulkrabek B.J. and Lim G.H. (2001). Sensitivity and specificity of somatic cell count and California Mastitis Test for identifying intramammary infection in early lactation. J. Dairy Sci. 84:2018-2024.

Sayd, T., Morzel, M., Chambon, C., Franck, M., Figwer, P., Larzul, C., Le Roy, P., Monin, G., Cherel, P. and Laville, E. (2006). Proteome analysis of the sarcoplasmic fraction of pig Semimembranosus muscle: Implications on meat colour development. Journal of Agricultural and Food Chemistry 54: 2732-2737.

Schaar, J. and Funke, H. (1986). Effect of subclinical mastitis on milk plasminogen and plasmin compared with that on sodium, antitrypsin and N-acetyl-β-D-glucosaminidase. Journal of Dairy Research 53: 515-528.

Schaefer, A. L., Cook, N., Tessaro, S.V., Deregt, D., Desroches, G., Dubeski, P.L., Tong, A.K.W. and Godson, D.L. (2004). Early detection and prediction of infection using infrared thermography. Can. J. Anim. Sci. 84:73–80.

Schaefer, A. L., Cook, N.J., Church, J.S., Basarab, J., Perry, B., Miller, C. and Tong, A.K.W. (2007). The use of infrared thermography as an early indicator of bovine respiratory disease complex in calves. Res. Vet. Sci. 83:376–384.

Schaefer, A. L., Jones, S.D.M., Murray, A.P., Sather, A.P. and Tong, A.K.W. (1989). Infrared thermography of pigs with known genotypes for stress susceptibility in relation to pork quality. Can. J. Anim. Sci. 69:491–495. 323

Schaer, D.J., Boretti, F.S., Schoedon, G. and Schaffner, A. (2002). Induction of the CD163-dependent haemoglobin uptake by macrophage as a novel anti-inflammatory action of glucocorticoids. British Journal of Haematology 119: 239-243.

Schalm, O.W. and Noorlander, B.S. (1957). Experiments and observations leading to development of the California Mastitis Test. JAVMA 130:199-204.

Schalm, O.W., Carroll, E.J. & Jain, N.C. (1971). Bovine Mastitis pp. 80-89. Philadelphia, PA: Lea&Febiger.

Schanbacher, F.L., Talhouk, R.S. and Murray, F.A. (1997). Biology and origin of bioactive peptides in milk. Livestock Production Science 50: 105-123.

Schmelzer, C.E.H., Schops, R., Ulbrich-Hofman, R., Neubert, R.H.H. and Raith, K. (2004). Mass spectrometric characterization of peptides derived by peptide cleavage of bovine beta-casein. Journal of Chromatography A. 1055: 87-92.

Schnyder, D. & Bagglioni, M. (1978). Secretion of lysosomal hydrolases by stimulated and nonstimulated macrophages. J. Exp. Med. 148:435.

Schoofs, L., Holman, G.M., Proost, P., Van Damme, J., Neven, H., Oudejans, R.C.H.M. and De Loof, A. (1993). Isolation, identification and synthesis of AKH-I4-10 from Locusta migratoria. General and Comparative Endocrinology 90: 364-371.

Schorlemmer, H.U., Davies, P., Hylton, W., Gugig, M. & Allison, A.C. (1977). The selective release of lysosomal acid hydrolases from mouse peritoneal macrophages by stimuli of chronic inflammation. J. Exp. Pathol. 58:313.

Schrader, M. and Schulz-Knappe, P. (2001). Peptidomics technologies for human body fluids. Trends in Biotechnology 19: S55-S60.

Schrader, M. and Selle, H. (2006). The process chain for peptidomic biomarker discovery. Disease Markers 22: 27-37.

Schroten, H., Hanisch, F.G., Plogmann, R., Hacker, J., Uhlenbruck, G., and Nobis-Bosch, R. and Wahn, V. (1992). Inhibition of adhesion of S-fimbriated Escherichia coli to bucal epithelial cells by human milk fat globule membrane components: function of mucins in the nonimmunoglobulin fraction. Infection and Immunity 60: 2893–2899.

Schukken, Y.H., Wilson, D.J., Welcome, F., Garrison-Tikofsky, L. and Gonzalez, R.N. (2003). Monitoring udder health and milk quality using somatic cell counts Vet. Res. 34: 579-596. 324

Schulltz, L.H. (1977). Somatic cell counting of milk in production testing programs as a mastitis control technique. J. Am. Vet. Med. Ass. 170: 1244-1246.

Schulte I., Tammen H., Selle H. and Schulz-Knappe P. (2005) Peptides in body fluids and tissues as markers of disease. Expert Review of Molecular Diagnostics. 5: 145-157.

Schulz-Knappe, P., Zucht, H.D., Heine, C., Jürgens, M., Hess, R. and Schrader, M. (2001). Peptidomics: the comprehensive analysis of peptides in complex biological mixtures. Comb. Chem. High Throughput Screen 4: 207–217.

Schwab, C.G., Bozak, C.K., Whitehouse, N.L. and Olson, V.M. (1992). Amino acid limitation and flow to the duodenum at four stages of lactation. 2. Extent of lysine limitation. Journal of Dairy Science 75: 3503-3518.

Scott, S. L., Schaefer, A.L., Tong, A.K.W. and Lacasse, P. (2000). Use of infrared thermography for early detection of mastitis in dairy cows. In Proc. CSAS Annual Meeting, Winnipeg, MB, Canada.

Seegers, H., Fourichon, C. and Beaudeau, F. (2003). Production effects related to mastitis and mastitis economics in dairy cattle herds. Veterinary Research 34: 475-491.

Seo, D. and Ginsburg, G.S. (2005). Genomic medicine: bringing biomarkers to clinical medicine. Current Opinion in Chemical Biology 9: 381-386.

Serhan, C.N., Chiang, N. and van Dyke, T.E. (2008). Resolving inflammation: dual anti-inflammatory and pro-resolution lipid mediators. Nat. Rev. Immunol. 8: 349-361.

Severin, S. and Wenshui, X. (2005). Milk biologically active components as nutraceuticals, Crit. Rev. Food Sci. Nutr. 45: 645–656 (Review)

Shainkin-Kestenbaum, R., Berlyne, G., Zimlichman, S., Sorin, H.R., Nyska, M. and Danon, A. (1991). Acute phase protein, serum amyloid A, inhibits IL-1- and TNF-induced fever and hypothalamic PGE2 in mice. Scand. J. Immunol. 34: 179-183.

Shamay, A., Homans, R., Fuerman, Y., Levin, I., Barash, H., Silanikove, N. and Mabjeesh, S.J. (2005). Expression of albumin in nonhepatic tissues and its synthesis by the bovine mammary gland. J. Dairy Sci 88: 569-576.

Sharif, A. and Muhammad, G. (2008). Somatic cell count as an indicator of udder health status under modern dairy production: A review. Pakistan Vet. J. 28:194-200.

Sharma, N. (2007). Alternative approach to control intramammary infection in dairy cows: A review. Asian Journal of Animal and Veterinary Advances 2 2:50-62. 325

Sharma, N., Maiti, S.K. and Roy, S. (2003). Role of Vitamin E in the control of mastitis in dairy cows. Vet. Practitioner 4:140-143.

Sharma, N., Pandey, V. and Sudhan, N.A. (2010). Comparison of some indirect screening tests for detection of subclinical mastitis in dairy cows. Bulgarian Journal of Veterinary Medicine 13:98-103.

Sheffield, C.L., Kamps-Holtzapplw, C., Deloach, J.R. and Stanker, L.H. (1994). Production and characterization of a monoclonal antibody against bovine haptoglobin and its use in an ELISA. Veterinary Immunology and Immunopathology 42: 171-183.

Sheldrake, R.F. and Hoare, R.J.T. (1981). The detection of mastitis in individual quarters using electrical conductivity or somatic cell concentration. N.Z. Vet J. 29:211-213.

Sheldrake, R.F., Hoare, R.J.T. and McGregor, G.D. (1983a). Lactation stage, parity and infection affecting somatic cells, electrical conductivity and serum albumin in milk. J. Dairy Sci 66:542-547.

Sheldrake, R.F., McGregor, G.D. and Hoare, R.J.T. (1983b). Somatic cell count, electrical conductivity and serum albumin concentration in detecting bovine mastitis. J. Dairy Sci 66:548-555.

Shen, Y., Zhao, R., Berger, S.J., Anderson, G.A., Rodriguez, N. and Smith, R.D. (2002). High- efficiency nanoscale liquid chromatography coupled on-line with mass spectrometry using nanoelectrospray ionization for proteomics. Anal Chem 74: 4235–4249.

Shook, G.E. (1989). Selection for disease resistance. J. Dairy Sci. 72: 1348–1353.

Silanikove, N., Shamay, A., Shinder, D. and Moran, A. (2000). Stress down regulates milk yield in cows by plasmin induced β-casein product that blocks K+ channels on the apical membranes. Life Sciences 67: 2201-2212.

Silanikove, N., Shapiro, F., Shamay, A. and Leitner, G. (2005). Role of xanthine oxidase, lactoperoxidase, and NO in the innate immune system of mammary secretion during active involution in dairy cows: manipulation with casein hydrolyzates. Free Radical Biology and Medicine 38: 1139- 1151.

Sipe, J.D. (2000). Serum amyloid A: from fibril to function. Current status. Amyloid-International Journal of Experimental and Clinical Investigation 7: 10-12.

Sipe, J.D., Gonnerman, W.A., Loose, L.D., Knapschaefer, G., Xie, W-J. and Franzblau, C. (1989). Direct binding enzyme-linked immunosorbent assay (ELISA) for serum amyloid A (SAA). Journal of Immunological Methods 125: 125-135. 326

Skinner, J.G., and Roberts, L. (1994). Haptoglobin as an indicator of infection in sheep. Veterinary Record 134: 33-36.

Skinner, J.G., Brown, R.A. and Roberts, L. (1991). Bovine haptoglobin response in clinically defined field conditions. Veterinary Record 128: 147–149.

Skogen, B., Borresen, A.L., Natvig, J.B., Berg, K. and Michaelsen, T.E. (1979). High-density lipoprotein as carrier for amyloid-related protein SAA in rabbit serum. Scandinavian Journal of Immunology 10: 39-45.

Skold, K., Svensson, M., Kaplan, A., Bjorkesten, L., Astrom, J. and Andren, P.E. (2002). A neuroproteomic approach to targeting neuropeptides in the brain. Proteomics 2: 447-454.

Slemmon, J.R., Wengenack, T.M. and Flood, D.G. (1997). Profiling of endogenous peptides as a tool for studying development and neurological disease. Biopolymers 43: 157-170.

Smacchi, E. and Gobbetti, M. (2000). Bioactive peptides in dairy products: synthesis and interaction with proteolytic enzymes. Food Microbiology 17: 129-141.

Smith, J.W. and Schultze, W.D. (1967). Variation in cell content of milk associated with time of sample collection. J. Dairy Sci. 50: 1083-1087.

Smolenski, G., Haines, S., Kwan, F. Y-S., Bond, J., Farr, V., Davis, S.R., Stelwagen, K. and Wheeler, T.T., (2007). Characterisation of host defence proteins in milk using a proteomic approach. Journal of Proteome Research 6: 207-215.

Solter, P.F., Hoffman, W.E., Hungerford, L., Siegel, J.P., Stdenis, S.H. and Dornier, J.L. (1991). Haptoglobin and ceruloplasmin as determinants of inflammation in dogs. American Journal of Veterinary Research 52: 1738–1742.

Somma, A., Ferranti, P., Addeo, F., Mauriello, R. and Chianese, L. (2008). Peptidomic approach based on combined capillary isoelectric focusing and mass spectrometry for the characterization of the plasmin primary products from bovine and water buffalo β-casein. Journal of Chromatography A. 1192: 294-300.

Sordillo, L.M. and Streicher, K.L. (2002). Mammary gland immunity and mastitis susceptibility. Journal of Mammary Gland Biology and Neoplasia 7:135-146.

Sordillo, L.M., Contreras, G.A. and Aitken, S.L. (2009). Metabolic factors affecting the inflammatory response of periparturient dairy cows. Animal Health Research Reviews 10: 53-63. 327

Soujala, L., Orro, T., Järvinen, H., Saatsi, J. and Pyörälä, S. (2008). Acute phase response in two consecutive experimentally induced E. coli intramammary infections in dairy cows. Acta Veterinaria Scandinavica 50: 18.

Soyeurt, H., Dehareng, F., Gengler, N., McParland, S., Wall, E., Berry, D.P., Coffey, M. and Dardenne, P. (2011) Mid-infrared prediction of bovine milk fatty acids across multiple breeds, production systems, and countries. J. Dairy Sci. 94: 1657–1667.

Spengler, B., Kirsch, D., Kaufmann, R. and Jaeger, E. (1992). Peptide sequencing by matrix assisted laser desorption mass spectrometry. Rapid Commun Mass Spectrom 6:105–108.

Spitsberg, V.L. and Gorewit, R.C. (2002). Isolation, purification and characterization of fatty-acid- binding protein from milk fat globule membrane: Effect of bovine growth hormone treatment. Pak. J. Nutr. 1: 43–48.

Spooner, R.L. and Miller, J.K. (1971). The measurement of haemoglobin reactive protein in ruminants as an aid to the diagnosis of acute inflammation. Vet. Rec. 88: 2–4.

Srinivas, S. and Prakash, V. (2010). Bioactive peptides from bovine milk alpha-casein: Isolation, characterization and multifunctional properties. International Journal of Peptide Research and Therapeutics. 16: 7-15.

Stasyk, T. and Huber, L.A. (2004). Zooming In: Fractionation strategies in proteomics. Proteomics 4: 3704-3716.

Steel, L.F., Trotter, M.G., Nakajima, P.B., Mattu, T.S., Gonye, G. and Block, T. (2002). Efficient and Specific Removal of Albumin from Human Serum Samples. Molecular and Cellular Proteomics 2: 262-270.

Stewart, M., Webster, J.R., Schaefer, A.L., Cook, N.J. and Scott, S.L. (2005). Infrared thermography as a non-invasive tool to study animal welfare. Anim. Welf. 14:319–325.

Stoughton, R.B. and Friend, S.H. (2005). How molecular profiling could revolutionize drug discovery. Nat. Rev. Drug Discov. 4: 345–350.

Suman, S.P., Faustman, C., Stamer, S.L. and Liebler, D.C. (2007). Proteomics of lipid oxidation- induced oxidation of procine and bovine oxymyoglobins. Proteomics 7: 628-640.

Swaisgood, H.E. (1982). Chemistry of milk protein. Developments in Dairy Chemistry 1: 1-59.

Swaisgood, H.E. (1993). Review and update of casein chemistry. Journal of Dairy Science 76: 3054- 3061. 328

Sweedler, J.V., Li, L.., Floyd, P. and Gilly, W. (2000). Mass spectrometric survey of peptides in cephalopods with an emphasis on the FMRFamide-related peptides. J. Exp. Biol. 203: 3565–3573.

Tammen, H., Kreipe, H., Hess, R., Kellman, M., Lehmann, U., Pich, A., Lamping, N., Schulz- Knappe, P., Zucht, H.D. and Lilischkis, R. (2003). Expression profiling of breast cancer cells by differential peptide display. Breast Cancer Research and Treatment 79: 83-93.

Tanaka, K., Waki, H., Ido, Y., Akita, S., Yoshida, Y. and Yoshida, T. (1988). Protein and polymer analysis up to m/z 100.000 by laser desorption time-of-flight mass spectrometry. Rapid Commun Mass Spectrom 2:151–153.

Tanwar, R. K., Vyas, S.K., Fakhruddin, & Singh, A.P. (2001). Comparative efficacy of various diagnostic tests in diagnosis of SCM in Rathi cows. In: Proceedings of Round Table Conference of the Indian Association for the Advancement of Veterinary Research (IAAVR) on Mastitis, 9-11th April 2001. pp. 161-163.

Tarukoski, P.H. (1966). Quantitative spectrophotometric determination of haptoglobin. Scandinav. J. Clin. & Lab. Investigation 18: 80-86.

Taverna, F., Negri, A., Piccinini, R., Zecconi, A., Nonnis, S., Ronchi, S. and Tedeschi G. (2007). Characterization of cell wall associated proteins of a Staphylococcus aureus isolated from bovine mastitis case by a proteomic approach. Veterinary Microbiology 119: 240-247.

Thielen, M.A., Mielenz, M., Hiss, S. and Sauerwein, H. (2005). Qualitative detection of haptoglobin mRNA in bovine and human blood leukocytes and bovine milk somatic cells. Vet. Med.-Czech 50: 515-520.

Tian, R., Jiang, X., Li, X., Jiang, X., Feng, S., Xu, S., Han, G., Ye, M. and Zou, H. (2006). Biological fingerprinting analysis of the interactome of a kinase inhibitor in human plasma by a chemiproteomic approach. Journal of Chromatography A 1134: 134-142.

Tian, R., Ren, L., Ma, H., Hu, L., Ye, M., Wu, R., Tian, Z., Liu, Z. and Zou, H. (2009). Selective enrichment of endogenous peptides by chemically modified porous nanoparticles for peptidome analysis. Journal of Chromatography A 1216: 1270-1278.

Tiemeyer, W., Stohrer, M. and Giesecke, D. (1984). Metabolites of nucleic acids in bovine milk. Journal of Dairy Science 67: 723-728.

Timms, J.F. and Cramer, R. (2008). Difference gel electrophoresis. Proteomics 8: 23-24. 329

Timms, L.L. & Schultz, L.H. (1985). N-acetyl-β-D-glycosamidase in milk and blood plasma during dry and early postpartum periods. Journal of Dairy Science 68:3367–3370.

Titov, V.Y., Kosenko, O.V., Fisinin, V.I. and Klimov, N.T. (2010). Content of nitric oxide metabolites in cow milk in health and mastitis. Russian Agricultural Sciences 36: 288-290.

Toba, Y., Takada, Y., Yamamura, J., Tanaka, M., Matsuoka, Y., Kawakami, H., Itabashi, A., Aoe, S. and Kumegawa, M. (2000). Milk basic protein: A novel protective function of milk against osteoporosis. Bone 27: 403-408.

Tomita, M. and Nishioka, T. (2005). Metabolomics: The Frontier of Systems Biology. Chapter 1: Overview Tokyo: Springer-Verlag, 1-6.

Travis, J. (1988). Structure, function, and control of neutrophil proteinases. Am J Med 84: 37–42.

Tyagi, S., Raghvendra, Singh, U., Kalra, T. and Munjal, K. (2010). Application of metabolomics-A systematic study of the unique chemical fingerprints: An overview. International Journal of Pharmaceutical Sciences Review and Research 3: 83-86.

Uchida, E., Katoh, N. and Takahashi, K. (1993). Appearance of haptoglobin in serum from cows at parturition. Journal of Veterinary Medical Science 55: 893–894.

Ullah, S., Ahmad, T., Bilal, M.Q., Zia-ur-rehman, Muhammad, G. and Rehman, S.U. (2005). The effect of severity of mastitis on protein and fat contents of buffalo milk. Pakistan Vet. J. 25: 1-4.

University of Nottingham (2008, September 18). Bovine Mastitis: Could A Vaccine Be On The Way? ScienceDaily. Retrieved March 13, 2011, from http://www.sciencedaily.com /releases/

Unlu, M. and Minden, J. (1997). Difference gel electrophoresis. The Protein Protocols Handbook 2nd Edition. Edited by: J. M. Walker © Humana Press Inc., Totowa, NJ.

Urieli-Shval, S., Cohen, P., Elsenberg, S. and Matzner, Y. (1998). Expression of serum amyloid A in histologically normal human tissues: predominant localization to the epithelium. J. Histochem. Cytochem. 46: 1377-1384.

Uttenweiler-Joseph, S., Moniatte, M., Lagueux, M., Van Dorsselaer, A., Hoffmann, J.A. and Bulet, P. (1998). Differential display of peptides induces during the immune response of Drosophila: a matrix- assisted laser desorption ionization time-of-flight mass spectrometry study. Proc. Natl. Acad. Sci. USA 95: 11342-11347. 330

Vaidyanathan, S., Kell, D.B. and Goodacre, R. (2002). Flow-injection electrospray ionization mass spectrometry of crude cell extracts for high-throughput bacterial identification. J. Am. Soc. Mass Spectrom. 13: 118–128.

Van Asbeck, B.S. (1990). Oxygen toxicity: Role of hydrogen peroxide and iron. Adv. Exp. Med. Biol. 264: 235–246.

Van den Bergh, G., Clerens, S., Vandesande, F. and Arckens, L. (2003). Reversed-phase high- performance liquid chromatography prefractionation prior to two-dimensional difference gel electrophoresis and mass spectrometry identifies new differentially expressed proteins between striate cortex of kitten and adult cat. Electrophoresis 24: 1471-1481.

Van Riel, J. and Olieman, C. (1995). Determination of caseinomacropeptide with capillary zone electrophoresis and its application to the detection and estimation of rennet whey solids in milk and buttermilk powder. Electrophoresis 16: 529-522.

Van Werven, T., Noordhuizen-Stassen, E.N., Daemen, A., Schukken, Y.H., Brand, A. and Burvenich, C. (1997). Pre-infection in vitro chemotaxis, phagocytosis, oxidative burst, and expression of CD11/CD18 receptors and their predictive capacity on the outcome of mastitis induced in dairy cows with Escherichia coli. J. Dairy Sci. 80:67–74.

Vanderghem, C., Blecker, C., Danthine, S., Deroanne, C., Haubruge, E., Guillonneau, F., De Pauw, E. and Francis, F. (2008). Proteome analysis of the bovine milk fat globule: Enhancement of membrane purification. International Dairy Journal 18: 885-893.

Verdi, R.J., Barbano, D.M., Dellavalle, M.E. and Senyk, G.F. (1987). Variability in true protein, casein, nonprotein nitrogen, and proteolysis in high and low somatic cell milks. Journal of Dairy Science 70: 230-242.

Verhaert, P., Uttenweiler-Joseph, S., de Vries, M., Loboda, A, Ens, W. and Standing, K.G. (2001). Matrix-assisted laser desorption/ionization quadrupole timeof-flight mass spectrometry: An elegant tool for peptidomics. Proteomics 1:118–131.

Viguier, C., Arora, S., Gilmartin, N., Welbeck, K. and O’Kennedy, R. (2009). Mastitis detection: current trends and future perspectives. Trends in Biotechnology 27:486-493.

Villanueva, J., Philip, J., Entenberg, D., Chaparro, C.A., Tanwar, M.K., Holland, E.C. and Tempst, P. (2004). Serum peptide profiling by magnetic particle-assisted, automated sample processing and MALDI-TOF mass spectrometry. Anal. Chem. 76:1560–1570. 331

Villanueva, J., Shaffer, D.R., Philip, J., Chaparro, C.A., Erdjument-Bromage, H., Olshen, A.B., Fleisher, M., Lilja, H., Brogi, E., Boyd, J., Sanchez-Carbayo, M., Holland, E.C., Cordon-Cardo, C., Scher, H.I. and Tempst, P. (2006). Differential exoprotease activities confer tumor-specific serum peptidome patterns. The Journal of Clinical Investigation 116: 271-284.

Viswanathan, S., Unlu, M. and Minden, J.S. (2006).Two-dimensional difference gel electrophoresis. Nat. Protoc 1: 1351-1358.

Vreugdenhill, A.C.E., Dentener, M.A., Snoek, A.M.P., Greve, J.W.M. and Buurman, W.A. (1999). Lipopolysaccharide binding protein and serum amyloid A secretion by human intestinal epithelial cells during the acute phase response. The Journal of Immunology 163: 2792-2798.

Wagener, F.A.D.T.G., Eggert, A., Boerman, O.C., Oyen, W.J., Verhofstad, A., Abraham, N.G., Adema, G., van Kooyk, Y., de Witte, T. and Figdor, C.G. (2001). Heme is a potent inducer of inflammation in mice and is counteracted by heme oxygenase. Blood 98:1802–1811.

Waldschmidt, M. and Rilling, S. (1973). Metabolite levels and enzyme activities in the bovine mammary gland in different stages of lactation. Journal of Dairy Research 40: 361-370.

Waller, K.P. (2000). Mammary gland immunology around parturition-Influence of stress, nutrition and genetics. Biology of the mammary gland. Advances in experimental medicine and biology 480: 231-245.

Wang, D. Y. (1960). Nature 188: 1109.

Wang, X., Hirmo, S., Willen, R. and Wadstrom, T. (2001). Inhibition of Helicobacter pylori infection by bovine milk glycoconjugates in a BALB/cA mouse model. Journal of Medical Microbiology 50: 430-435.

Wang, Y., Jen-Fu, C. & Qing-Yu, H. (2007). Bioinformatic application in proteomic research on biomarker discovery and drug target validation. Current Bioinformatics 2: 11-20.

Ward, P.P., Uribe-Luna, S. and Conneely, O.M. (2002). Lactoferrin and host defense. Biochem Cell Biol 80: 95-102.

Washburn, M.P., Wolters, D. and Yates, J.R., (2001). Large-scale analysis of the yeast proteome by multidimensional protein identification technology. Nat Biotechnol. 19: 242-247.

Wattiauz, M.A. and Karg, K.L. (2004). Protein level for alfalfa and corn silage-based diets: II. Nitrogen balance and manure characteristics. Journal of Dairy Science 87: 3492-3502. 332

Weber, A., Weber, A.T., McDonald, T.L. and Larson, M.A. (2005). Staphylococcus aureus lipotechoic acid induces differential expression of bovine serum amyloid A3 (SAA3) by mammary epithelial cells: Implications for early diagnosis of mastitis. Veterinary Immunology and Immunopathology 109: 79-83.

Wedholm, A., Møller, H.S., Lindmark-Månsson, H., Rasmussen, M.D., Andrén, A. and Larsen, L.B. (2008). Identification of peptides in milk as a result of proteolysis at different levels of somatic cell counts using LC MALDI MS/MS detection. Journal of Dairy Research 75: 76-83.

Weekes, J., Wheeler, C.H., Yan, J.X., Weil, J., Eschenhagen, T., Scholtysik, G. and Dunn, M.J. (1999). Bovine dilated cardiomyopathy: Proteomic analysis of an animal model of human dilated cardiomyopathy. Electrophoresis 20: 898-906.

Weiss, J.M. (1989). Tissue destruction by neutrophils. New Engl. J. Med. 320: 365-376.

Wenz, J.R., Fox, L.K., Muller, F.J., Rinaldi, M., Zeng, R. and Bannerman, D.D. (2010). Factors associated with concentrations of select cytokine and acute phase proteins in dairy cows with naturally occurring clinical mastitis. Journal of Dairy Science 93: 2458-2470.

Wesen, D.P., Luedecke, L.O. and Forster, T.L. (1967). Relationship between California Mastitis Test reaction and bacteriological analyses of stripping samples. J. Dairy Sci 51:679-684.

Whitfield, P.D., Noble, P.J.M., Major, H., Beynon, R.J., Burrow, R., Freeman, A.I. and German, A.J. (2005). Metabolomics: as a diagnostic tool for hepatology: validation in a naturally occurring canine model. Metabolomics 1: 215-225.

Wilkins, M.R., Sanchez, J-C., Gooley, A.A., Appel, R.D., Humphery-Smith, I., Hochstrasser, D.F. and Williams, K.L. (1995). Progress with proteome projects: why all proteins expressed by a genome should be identified and how to do it. Biotechnol. Genet. Eng. Rev. 13: 19–50.

Williams, K.O. and Higgs, G.A. (1988). Eicosanoids and inflammation. The Journal of Pathology 156: 101-110.

Wishart, D.S. (2008). Quantitative metabolomics using NMR. Trends in Analytical Chemistry 27: 228-237.

Wolfe, R. R., Sharma, S. C. & Ward, G. E. (1972). A method for electronic detection of bovine mastitis. Transactions of the American Society of Agricultural Engineers 15: 500-503.

Wong, C.W., Seow, H.F., Liu, A.H., Husband, A.J., Smithers, G.W. & Watson, D.L. (1996). Modulation of immune responses by bovine beta-casein. Immunology and Cell Biology 74: 323-329. 333

Wong, N.P., Physical properties of milk. (1988) Page 409 in Fundamentals of dairy chemistry. 3rd Ed. N.P. Wong, ed Van Nostrand Reinhold Co., New York, NY.

Woolford, M.W. & Williamson, J. H. (1982). The electrical conductivity of milk as a diagnostic tool of subclinical mastitis. Dairy Production from Pasture pp. 114-115 (Eds K. L. Macmillan and V. K. Taufa) Hamilton, New Zealand: Clarke & Matheson Ltd.

Woolford, M.W., Williamson, J.H. and Henderson, H.V. (1998). Change in electrical conductivity and somatic cell count between milk fractions from quarters subclinically infected with particular mastitis pathogens. J. Dairy Res. 65:187-198.

Wu, Y., Chen, Y., Wang, B., Wang, H., Yuan, F. and Zhao, G. (2009). 2DGE-coomassie brilliant blue staining used to differentiate pasteurized milk from reconstituted milk. Health 1: 146-151.

Xie, B., Tassi, E., Swift, M.R., McDonnell, K., Bowden, E.T., Wang, S., Ueda, Y., Tomita, Y., Riegel, A.T. and Wellstein, A. (2006). Identification of the fibroblast growth factor (FGF)-interacting domain in a secreted FGF-binding protein by phage display. J Biol. Chem. 281: 1137-1144.

Xie, Y., Li, Y., Zhang, Q., Stiller, M.J., Albert Wang, C.-L. and Streilein, J.W. (2000). Haptoglobin is a natural regulator of Langerhans cell function in the skin. Journal of Dermatological Science. 24: 25- 37.

Xu, L., Badolato, R., Murphy, W.J., Longo, D.L., Anver, M., Hale, S., Oppenheim, J.J. and Wang, J.M. (1995). A novel biologic function of serum amyloid A. Induction of T lymphocyte migration and adhesion. J Immunol 155: 1184–1190.

Yamada, M., Murakami, K., Wallingford, J.C. and Yuki Y. (2002). Identification of low-abundance proteins of bovine colostral and mature milk using two-dimensional electrophoresis followed by microsequencing and mass spectrometry. Electrophoresis 23: 1153–1160.

Yamada, M., Murakami, K., Wallingford, J.C. and Yuki, Y. (2002). Identification of low-abundance proteins of bovine colostral and mature milk using two-dimensional electrophoresis followed microsequencing and mass spectrometry. Electrophoresis 23: 1153-1160.

Yamada, T., Nomata, Y., Sugita, O. and Okada, M. (1993). A rapid method for measuring serum amyloid A protein by latex agglutination nephelometric immunoassay. Ann Clin. Biochem. 30: 72-76.

Yamamoto, M. (1982). Personal communication 334

Yamamoto, M., Oohashi, T., Katoh, N. and Oikawa, S. (2000). Increased serum concentration of apolipoprotein C-III and its greater distribution of chylomicrons than to the high density lipoprotein fraction in a calf with hyperlipidemia. Journal of Veterinary Medical Science 62: 1033–1039.

Yamauchi, K., Tomita, M., Giehl, T.J. and Ellison, R.T. (1993). Antibacterial activity of lactoferrin and a pepsin-derived lactoferrin peptide fragment. Infection and Immunity 61: 719-728.

Yang, F., Haile, D.J., Berger, F.G., Herbert, D.C., Van Beveren, E. & Ghio, A.J. (2003). Haptoglobin reduces lung injury associated with exposure to blood. Am J Physiol Lung Cell Mol Physiol 284: L402–L409

Yang, Y.X., Zhao, X.X. and Zhang, Y. (2009). Proteomic analysis of mammary tissues from healthy cows and clinical mastitic cows for identification of disease-related proteins. Veterinary Research Communications 33: 295-303.

Yong, Z., Xin, Y.Y., and Xu, Z.X., (2009). Comparative proteome analysis of changes in clinical healthy and mastitic cows mammary gland expression protein. Scientia Agricultura Sinica 42: 1442- 1446.

Yoshino, K., Katoh, N., Takahashi, K. and Yuasa, A. (1993). Possible involvement of protein kinase C with induction of haptoglobin in cows by treatment with dexamethasone and by starvation. American Journal of Veterinary Research 54: 689–694.

Yoshioka, M., Watanabe, A., Shimada, N., Murata, H., Yokomizo, Y. and Nakajima, Y. (2002). Regulation of haptoglobin secretion by recombinant bovine cytokines in primary cultured bovine hepatocytes. Domestic Animal Endocrinology 23: 425–433.

Young, C.R., Eckersall, P.D., Saini, P.K. and Stanker, L.H. (1995). Validation of immunoassays for bovine haptoglobin. Veterinary Immunology and Immunopathology 49: 1-13.

Yuan, X.L. and Desiderio, D.M. (2005). Human cerebrospinal fluid peptidomics. Journal of Mass Spectrometry 40: 176-181.

Zank, W. and Schlatterer, B., (1998). Assessment of subacute mammary inflammation by soluble biomarkers in comparison to somatic cell counts in quarter milk samples from dairy cows. Journal of Veterinary Medicine Series A 45: 41-51.

Zecconi, A. and Piccinini, R. Intramammary infections: epidemiology and diagnosis. In: Kaske, M., Scholz, H. and Höltershinken, M. (eds) Recent Developments and Perspectives in Bovine Medicine. Keynote Lectures. XIIth World Buiatrics Congress 18–23 August, 2002. Hannover, Germany. pp. 346–359. 335

Zeng, R., Bequette, B.J., Vinyard, B.T. and Bannerman, D.D. (2009). Determination of milk and blood concentrations of lipopolysaccharide-binding protein in cows with naturally acquired subclinical and clinical mastitis. Journal of Dairy Science 92: 980-989.

Zimlichman, S., Danon, A., Nathan, I., Mozes, G. and Shainkin-Kestenbaumm, R. . (1990). Serum amyloid A, an acute phase protein, inhibits platelet activation. J. Lab. Clin. Med. 116: 180–186.

Zipser, R.D. and Laffi, G. (1985). Prostaglandins, thromboxanes and leukotrienes in clinical medicine. West J. Med. 143: 485-497.

Zuo, X. and Speicher, D. W. (2002). Comprehensive analysis of complex proteomes using microscale solution isoelectricfocusing prior to narrow pH range two-dimensional electrophoresis. Proteomics 2: 58-68.

Zürbig, P., Decramer, S., Dakna, M., Jantos, J., Good, D.M., Coon, J.J., Bandin, F., Mischak, H., Bascandas, J.L. and Schanstra, J.P. (2009). The human urinary proteome reveals high similarity between kidney aging and chronic kidney disease. Proteomics 9: 2108-2117.